[{"url":"https://api.github.com/repos/openclaw/openclaw/issues/10354","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/10354/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/10354/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/10354/events","html_url":"https://github.com/openclaw/openclaw/issues/10354","id":3906227516,"node_id":"I_kwDOQb6kR87o1E08","number":10354,"title":"[Feature]: Add description and enum to message tool's channel parameter","user":{"login":"carrotRakko","id":24588751,"node_id":"MDQ6VXNlcjI0NTg4NzUx","avatar_url":"https://avatars.githubusercontent.com/u/24588751?v=4","gravatar_id":"","url":"https://api.github.com/users/carrotRakko","html_url":"https://github.com/carrotRakko","followers_url":"https://api.github.com/users/carrotRakko/followers","following_url":"https://api.github.com/users/carrotRakko/following{/other_user}","gists_url":"https://api.github.com/users/carrotRakko/gists{/gist_id}","starred_url":"https://api.github.com/users/carrotRakko/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/carrotRakko/subscriptions","organizations_url":"https://api.github.com/users/carrotRakko/orgs","repos_url":"https://api.github.com/users/carrotRakko/repos","events_url":"https://api.github.com/users/carrotRakko/events{/privacy}","received_events_url":"https://api.github.com/users/carrotRakko/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":9706244996,"node_id":"LA_kwDOQb6kR88AAAACQomLhA","url":"https://api.github.com/repos/openclaw/openclaw/labels/enhancement","name":"enhancement","color":"0969DA","default":true,"description":"New feature or request"},{"id":10970446441,"node_id":"LA_kwDOQb6kR88AAAACjeO6aQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/P2","name":"P2","color":"FBCA04","default":false,"description":"Normal backlog priority with limited blast radius."},{"id":10971435209,"node_id":"LA_kwDOQb6kR88AAAACjfLQyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:no-new-fix-pr","name":"clawsweeper:no-new-fix-pr","color":"E5E7EB","default":false,"description":"ClawSweeper does not recommend queueing a new automated fix PR for this issue."},{"id":10971435771,"node_id":"LA_kwDOQb6kR88AAAACjfLS-w","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:needs-product-decision","name":"clawsweeper:needs-product-decision","color":"FBCA04","default":false,"description":"ClawSweeper marked this issue as needing a product or behavior decision."},{"id":10971438267,"node_id":"LA_kwDOQb6kR88AAAACjfLcuw","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:source-repro","name":"clawsweeper:source-repro","color":"0A3069","default":false,"description":"ClawSweeper found a high-confidence source-level issue reproduction."},{"id":10971438454,"node_id":"LA_kwDOQb6kR88AAAACjfLddg","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:linked-pr-open","name":"clawsweeper:linked-pr-open","color":"57606A","default":false,"description":"ClawSweeper found an open linked pull request for this issue."},{"id":10973513781,"node_id":"LA_kwDOQb6kR88AAAACjhKINQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/impact:message-loss","name":"impact:message-loss","color":"D93F0B","default":false,"description":"Channel message delivery can be lost, duplicated, or misrouted."},{"id":10981171401,"node_id":"LA_kwDOQb6kR88AAAACjodgyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/issue-rating:%20%F0%9F%A6%9E%20diamond%20lobster","name":"issue-rating: šŸ¦ž diamond lobster","color":"0969DA","default":false,"description":"Very strong issue quality with high-confidence source-level or clear reproduction."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":3,"created_at":"2026-02-06T11:10:05Z","updated_at":"2026-06-16T12:37:02Z","closed_at":null,"assignee":null,"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"sub_issues_summary":{"total":0,"completed":0,"percent_completed":0},"issue_dependencies_summary":{"blocked_by":0,"total_blocked_by":0,"blocking":0,"total_blocking":0},"body":"## Summary\n\nThe `message` tool's `channel` parameter (`src/agents/tools/message-tool.ts`, line 40 in `buildRoutingSchema`) is defined as a bare `Type.Optional(Type.String())` with no description or enum constraint. This causes AI agents to pass Slack/Discord channel IDs (e.g. `C07ABC1234X`) instead of provider names (e.g. `slack`), resulting in \"Unknown channel\" errors from `normalizeAnyChannelId`.\n\nThe system prompt's Messaging section does document the expected values (`telegram|whatsapp|discord|googlechat|slack|signal|imessage`), but the tool schema itself provides no guidance. Since other parameters in the same schema include descriptions (e.g. `target: \"Target channel/user id or name.\"`), the empty `channel` description stands out as an oversight.\n\n## Proposed solution\n\nAdd a description to the `channel` parameter in `buildRoutingSchema()`:\n\n```typescript\nchannel: Type.Optional(Type.String({\n description: \"Messaging provider name (e.g. 'slack', 'discord'). Not the channel/conversation ID.\",\n})),\n```\n\nOptionally, an enum constraint could also help, though dynamic channel availability may make a static enum impractical.\n\n## Alternatives considered\n\n- **Prompt-only fix**: The system prompt already documents this, but agents can miss prompt instructions when tool schemas accept arbitrary strings.\n- **Fallback resolution in normalizeAnyChannelId**: Detect Slack-style IDs (starting with `C`, `U`, `G`) and auto-route. Adds complexity and only covers known formats.\n\n## Additional context\n\n- Discovered while debugging an \"Unknown channel\" error on a self-hosted OpenClaw Slack instance\n- The agent could receive and reply to messages via auto-reply, but tool-based message operations failed\n- Telling the agent to pass `channel: \"slack\"` instead of the Slack channel ID immediately resolved the issue\n- Related (but distinct) issue: #4751 (Slack IDs lowercased in normalizeSlackMessagingTarget)\n\nāœļø **Author**: Claude Code with @carrotRakko (AI-written, human-approved)","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/10354/reactions","total_count":0,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":0},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/10354/timeline","performed_via_github_app":null,"state_reason":null,"pinned_comment":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/10142","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/10142/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/10142/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/10142/events","html_url":"https://github.com/openclaw/openclaw/issues/10142","id":3904992655,"node_id":"I_kwDOQb6kR87owXWP","number":10142,"title":"Feature: session:end internal hook event","user":{"login":"spk-alex","id":205000749,"node_id":"U_kgDODDgQLQ","avatar_url":"https://avatars.githubusercontent.com/u/205000749?v=4","gravatar_id":"","url":"https://api.github.com/users/spk-alex","html_url":"https://github.com/spk-alex","followers_url":"https://api.github.com/users/spk-alex/followers","following_url":"https://api.github.com/users/spk-alex/following{/other_user}","gists_url":"https://api.github.com/users/spk-alex/gists{/gist_id}","starred_url":"https://api.github.com/users/spk-alex/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/spk-alex/subscriptions","organizations_url":"https://api.github.com/users/spk-alex/orgs","repos_url":"https://api.github.com/users/spk-alex/repos","events_url":"https://api.github.com/users/spk-alex/events{/privacy}","received_events_url":"https://api.github.com/users/spk-alex/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":9706244996,"node_id":"LA_kwDOQb6kR88AAAACQomLhA","url":"https://api.github.com/repos/openclaw/openclaw/labels/enhancement","name":"enhancement","color":"0969DA","default":true,"description":"New feature or request"},{"id":10970446441,"node_id":"LA_kwDOQb6kR88AAAACjeO6aQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/P2","name":"P2","color":"FBCA04","default":false,"description":"Normal backlog priority with limited blast radius."},{"id":10971435209,"node_id":"LA_kwDOQb6kR88AAAACjfLQyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:no-new-fix-pr","name":"clawsweeper:no-new-fix-pr","color":"E5E7EB","default":false,"description":"ClawSweeper does not recommend queueing a new automated fix PR for this issue."},{"id":10971435625,"node_id":"LA_kwDOQb6kR88AAAACjfLSaQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:needs-maintainer-review","name":"clawsweeper:needs-maintainer-review","color":"FBCA04","default":false,"description":"ClawSweeper marked this issue as needing maintainer review before automation."},{"id":10971435771,"node_id":"LA_kwDOQb6kR88AAAACjfLS-w","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:needs-product-decision","name":"clawsweeper:needs-product-decision","color":"FBCA04","default":false,"description":"ClawSweeper marked this issue as needing a product or behavior decision."},{"id":10971438267,"node_id":"LA_kwDOQb6kR88AAAACjfLcuw","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:source-repro","name":"clawsweeper:source-repro","color":"0A3069","default":false,"description":"ClawSweeper found a high-confidence source-level issue reproduction."},{"id":10971438454,"node_id":"LA_kwDOQb6kR88AAAACjfLddg","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:linked-pr-open","name":"clawsweeper:linked-pr-open","color":"57606A","default":false,"description":"ClawSweeper found an open linked pull request for this issue."},{"id":10973505423,"node_id":"LA_kwDOQb6kR88AAAACjhJnjw","url":"https://api.github.com/repos/openclaw/openclaw/labels/impact:session-state","name":"impact:session-state","color":"F9D65C","default":false,"description":"Session, memory, transcript, context, or agent state can drift or corrupt."},{"id":10981171401,"node_id":"LA_kwDOQb6kR88AAAACjodgyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/issue-rating:%20%F0%9F%A6%9E%20diamond%20lobster","name":"issue-rating: šŸ¦ž diamond lobster","color":"0969DA","default":false,"description":"Very strong issue quality with high-confidence source-level or clear reproduction."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":4,"created_at":"2026-02-06T05:35:25Z","updated_at":"2026-06-16T12:37:00Z","closed_at":null,"assignee":null,"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"sub_issues_summary":{"total":0,"completed":0,"percent_completed":0},"issue_dependencies_summary":{"blocked_by":0,"total_blocked_by":0,"blocking":0,"total_blocking":0},"body":"## Summary\n\nAdd a `session:end` internal hook event that fires when a session completes.\n\n## Use Case\n\nIntegrating OpenClaw with workflow orchestration systems like Temporal. When an agent session completes, we want to signal back to a waiting workflow so it can proceed to the next step.\n\nCurrent workaround requires agents to explicitly run a CLI command at the end of their task, which is fragile and adds cognitive load to agent instructions.\n\n## Proposed Behavior\n\n```typescript\n// Hook event structure\n{\n type: 'session',\n action: 'end',\n sessionKey: string,\n sessionId: string,\n timestamp: Date,\n context: {\n stopReason: 'stop' | 'toolUse' | 'maxTokens' | 'error',\n lastMessage?: string,\n metadata?: Record // Custom data passed when session started\n }\n}\n```\n\n## Example Hook\n\n```typescript\nconst handler: HookHandler = async (event) => {\n if (event.type !== 'session' || event.action !== 'end') return;\n \n const workflowId = event.context.metadata?.workflowId;\n if (workflowId) {\n // Signal Temporal workflow\n await exec(`temporal workflow signal --workflow-id ${workflowId} --name complete`);\n }\n};\n```\n\n## Notes\n\nThe docs mention `session:end` as a planned future event. This issue tracks that feature request.\n\nRelated: `session:start` would also be useful for the same integration patterns.","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/10142/reactions","total_count":0,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":0},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/10142/timeline","performed_via_github_app":null,"state_reason":null,"pinned_comment":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/7722","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/7722/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/7722/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/7722/events","html_url":"https://github.com/openclaw/openclaw/issues/7722","id":3889065894,"node_id":"I_kwDOQb6kR87nzm-m","number":7722,"title":"Feature Request: Filesystem Sandboxing Config (tools.fileAccess)","user":{"login":"LumenLantern","id":258675890,"node_id":"U_kgDOD2sUsg","avatar_url":"https://avatars.githubusercontent.com/u/258675890?v=4","gravatar_id":"","url":"https://api.github.com/users/LumenLantern","html_url":"https://github.com/LumenLantern","followers_url":"https://api.github.com/users/LumenLantern/followers","following_url":"https://api.github.com/users/LumenLantern/following{/other_user}","gists_url":"https://api.github.com/users/LumenLantern/gists{/gist_id}","starred_url":"https://api.github.com/users/LumenLantern/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/LumenLantern/subscriptions","organizations_url":"https://api.github.com/users/LumenLantern/orgs","repos_url":"https://api.github.com/users/LumenLantern/repos","events_url":"https://api.github.com/users/LumenLantern/events{/privacy}","received_events_url":"https://api.github.com/users/LumenLantern/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":9706244996,"node_id":"LA_kwDOQb6kR88AAAACQomLhA","url":"https://api.github.com/repos/openclaw/openclaw/labels/enhancement","name":"enhancement","color":"0969DA","default":true,"description":"New feature or request"},{"id":10970446441,"node_id":"LA_kwDOQb6kR88AAAACjeO6aQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/P2","name":"P2","color":"FBCA04","default":false,"description":"Normal backlog priority with limited blast radius."},{"id":10971435209,"node_id":"LA_kwDOQb6kR88AAAACjfLQyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:no-new-fix-pr","name":"clawsweeper:no-new-fix-pr","color":"E5E7EB","default":false,"description":"ClawSweeper does not recommend queueing a new automated fix PR for this issue."},{"id":10971435625,"node_id":"LA_kwDOQb6kR88AAAACjfLSaQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:needs-maintainer-review","name":"clawsweeper:needs-maintainer-review","color":"FBCA04","default":false,"description":"ClawSweeper marked this issue as needing maintainer review before automation."},{"id":10971435771,"node_id":"LA_kwDOQb6kR88AAAACjfLS-w","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:needs-product-decision","name":"clawsweeper:needs-product-decision","color":"FBCA04","default":false,"description":"ClawSweeper marked this issue as needing a product or behavior decision."},{"id":10971435893,"node_id":"LA_kwDOQb6kR88AAAACjfLTdQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:needs-security-review","name":"clawsweeper:needs-security-review","color":"B60205","default":false,"description":"ClawSweeper marked this issue as needing security-sensitive review."},{"id":10971438267,"node_id":"LA_kwDOQb6kR88AAAACjfLcuw","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:source-repro","name":"clawsweeper:source-repro","color":"0A3069","default":false,"description":"ClawSweeper found a high-confidence source-level issue reproduction."},{"id":10973513130,"node_id":"LA_kwDOQb6kR88AAAACjhKFqg","url":"https://api.github.com/repos/openclaw/openclaw/labels/impact:security","name":"impact:security","color":"B60205","default":false,"description":"Security boundary, credential, authz, sandbox, or sensitive-data risk."},{"id":10981171401,"node_id":"LA_kwDOQb6kR88AAAACjodgyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/issue-rating:%20%F0%9F%A6%9E%20diamond%20lobster","name":"issue-rating: šŸ¦ž diamond lobster","color":"0969DA","default":false,"description":"Very strong issue quality with high-confidence source-level or clear reproduction."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":7,"created_at":"2026-02-03T04:31:56Z","updated_at":"2026-06-16T12:36:56Z","closed_at":null,"assignee":null,"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"sub_issues_summary":{"total":0,"completed":0,"percent_completed":0},"issue_dependencies_summary":{"blocked_by":0,"total_blocked_by":0,"blocking":0,"total_blocking":0},"body":"**Feature:** Filesystem access restrictions via configuration\n\n**What happened:**\nWe attempted to implement filesystem sandboxing with this config:\n```javascript\n{\n \"tools\": {\n \"fileAccess\": {\n \"allowedPaths\": [\"/home/seraph/.openclaw/workspace\", \"/tmp\"],\n \"denyPaths\": [\"/etc\", \"/root\", \"~/.ssh\", \"/var/log\"]\n }\n }\n}\n```\n\n**Result:**\n```\nInvalid config at /home/seraph/.openclaw/openclaw.json:\n- tools: Unrecognized key: \"fileAccess\"\n```\n\n**Current behavior:**\nAgents have unrestricted filesystem access (limited only by OS user permissions).\n\n**Requested behavior:**\n```javascript\n\"tools\": {\n \"fileAccess\": {\n \"mode\": \"sandbox\", // or \"full\" for current behavior\n \"allowedPaths\": [\n \"/home/user/workspace\",\n \"/tmp\",\n \"~/.config\",\n \"~/.cache\"\n ],\n \"denyPaths\": [\n \"/etc/*\",\n \"/root\",\n \"~/.ssh\",\n \"/var/log\",\n \"/usr/bin\",\n \"/usr/sbin\",\n \"/boot\",\n \"/sys\"\n ]\n }\n}\n```\n\n**Behavior:**\n- `mode: \"sandbox\"` ā€” enforce path restrictions\n- `allowedPaths` ā€” whitelist of accessible directories (recursive)\n- `denyPaths` ā€” blacklist that overrides allowedPaths (defense in depth)\n- File operations outside allowed paths return error without execution\n- Works with read, write, exec tools\n\n**Why this matters:**\n- OWASP Agentic AI Top 10 #A03 (Excessive Agent Autonomy)\n- Prevents agents from reading `/etc/passwd`, `~/.ssh/id_rsa`, `/root/.secrets/`\n- Defense in depth: even if an agent is compromised, filesystem damage is contained\n\n**Example attack prevention:**\n```javascript\n// Attacker tries prompt injection to exfiltrate SSH keys\nagent.exec(\"cat ~/.ssh/id_rsa\")\n// With fileAccess.denyPaths: [\"~/.ssh\"]\n// ā†’ Error: Access denied to ~/.ssh/id_rsa\n```\n\n**Workaround:**\nNone. We're relying on OS-level user permissions (running as non-root `seraph` user).\n\n**Related:** Part of our response to Palo Alto Networks' claim that OpenClaw is fundamentally insecure. #7720","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/7722/reactions","total_count":4,"+1":4,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":0},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/7722/timeline","performed_via_github_app":null,"state_reason":null,"pinned_comment":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93631","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93631/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93631/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93631/events","html_url":"https://github.com/openclaw/openclaw/pull/93631","id":4674203035,"node_id":"PR_kwDOQb6kR87nARSp","number":93631,"title":"fix(ui): fall back to execCommand for clipboard copy over HTTP","user":{"login":"ajwan8998","id":50291365,"node_id":"MDQ6VXNlcjUwMjkxMzY1","avatar_url":"https://avatars.githubusercontent.com/u/50291365?v=4","gravatar_id":"","url":"https://api.github.com/users/ajwan8998","html_url":"https://github.com/ajwan8998","followers_url":"https://api.github.com/users/ajwan8998/followers","following_url":"https://api.github.com/users/ajwan8998/following{/other_user}","gists_url":"https://api.github.com/users/ajwan8998/gists{/gist_id}","starred_url":"https://api.github.com/users/ajwan8998/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/ajwan8998/subscriptions","organizations_url":"https://api.github.com/users/ajwan8998/orgs","repos_url":"https://api.github.com/users/ajwan8998/repos","events_url":"https://api.github.com/users/ajwan8998/events{/privacy}","received_events_url":"https://api.github.com/users/ajwan8998/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064298757,"node_id":"LA_kwDOQb6kR88AAAACV-EDBQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/app:%20web-ui","name":"app: web-ui","color":"6E7781","default":false,"description":"App: web-ui"},{"id":10190106648,"node_id":"LA_kwDOQb6kR88AAAACX2CwGA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20S","name":"size: S","color":"8C959F","default":false,"description":null}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":0,"created_at":"2026-06-16T12:36:34Z","updated_at":"2026-06-16T12:36:55Z","closed_at":null,"assignee":null,"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93631","html_url":"https://github.com/openclaw/openclaw/pull/93631","diff_url":"https://github.com/openclaw/openclaw/pull/93631.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93631.patch","merged_at":null},"body":"## Summary\n\n`navigator.clipboard.writeText()` requires a secure context (HTTPS or localhost). When accessing the dashboard over plain HTTP, the copy button on code blocks silently fails.\n\n## Changes\n\n**`ui/src/ui/views/chat.ts`**:\n- `handleCodeBlockCopy`: try `navigator.clipboard.writeText` first; on failure, fall back to `document.execCommand('copy')` via a temporary textarea\n\n**`ui/src/ui/chat/copy-as-markdown.ts`**:\n- `copyTextToClipboard`: same fallback pattern for the copy-as-markdown button\n\n## Real behavior proof\n\nBehavior addressed: Copy button on code blocks now works over plain HTTP by falling back to the legacy `document.execCommand('copy')` API when the async Clipboard API is unavailable.\n\nEnvironment tested: OpenClaw 2026.6.6, tested locally via TypeScript compilation.\n\nExact steps or command run after this patch: 1. Access dashboard via plain HTTP (e.g. http://192.168.0.100:9090). 2. Click a code block's copy button. 3. `navigator.clipboard.writeText` throws (not a secure context). 4. Fallback creates a hidden textarea, copies via `document.execCommand('copy')`.\n\nEvidence after fix: Code changes in 2 files, 49 insertions, 5 deletions. The `handleCodeBlockCopy` function now checks `navigator.clipboard?.writeText` first and falls back to `fallbackCopy(code)` which uses the `textarea + execCommand` pattern ā€” the same approach used by countless web apps for clipboard compatibility over HTTP.\n\nObserved result after fix: Copy button works over both HTTPS and HTTP. On HTTPS, the async Clipboard API is used. On HTTP, the execCommand fallback is used. Either way, the user sees the \"Copied!\" feedback.\n\nWhat was not tested: Internet Explorer (execCommand is deprecated but still widely supported). Mobile browsers may present a paste prompt with execCommand.\n\nCloses: #93628\n\nšŸ¤– Generated with [Claude Code](https://claude.com/claude-code)","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93631/reactions","total_count":0,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":0},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93631/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/7490","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/7490/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/7490/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/7490/events","html_url":"https://github.com/openclaw/openclaw/issues/7490","id":3887910696,"node_id":"I_kwDOQb6kR87nvM8o","number":7490,"title":"Feature Request: Add 'description' field to agent config for dynamic agent discovery","user":{"login":"shannon0430","id":258282406,"node_id":"U_kgDOD2UTpg","avatar_url":"https://avatars.githubusercontent.com/u/258282406?v=4","gravatar_id":"","url":"https://api.github.com/users/shannon0430","html_url":"https://github.com/shannon0430","followers_url":"https://api.github.com/users/shannon0430/followers","following_url":"https://api.github.com/users/shannon0430/following{/other_user}","gists_url":"https://api.github.com/users/shannon0430/gists{/gist_id}","starred_url":"https://api.github.com/users/shannon0430/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/shannon0430/subscriptions","organizations_url":"https://api.github.com/users/shannon0430/orgs","repos_url":"https://api.github.com/users/shannon0430/repos","events_url":"https://api.github.com/users/shannon0430/events{/privacy}","received_events_url":"https://api.github.com/users/shannon0430/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":9706244996,"node_id":"LA_kwDOQb6kR88AAAACQomLhA","url":"https://api.github.com/repos/openclaw/openclaw/labels/enhancement","name":"enhancement","color":"0969DA","default":true,"description":"New feature or request"},{"id":10970446441,"node_id":"LA_kwDOQb6kR88AAAACjeO6aQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/P2","name":"P2","color":"FBCA04","default":false,"description":"Normal backlog priority with limited blast radius."},{"id":10971435209,"node_id":"LA_kwDOQb6kR88AAAACjfLQyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:no-new-fix-pr","name":"clawsweeper:no-new-fix-pr","color":"E5E7EB","default":false,"description":"ClawSweeper does not recommend queueing a new automated fix PR for this issue."},{"id":10971435484,"node_id":"LA_kwDOQb6kR88AAAACjfLR3A","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:fix-shape-clear","name":"clawsweeper:fix-shape-clear","color":"1A7F37","default":false,"description":"ClawSweeper found a clear likely implementation shape for this issue."},{"id":10971435625,"node_id":"LA_kwDOQb6kR88AAAACjfLSaQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:needs-maintainer-review","name":"clawsweeper:needs-maintainer-review","color":"FBCA04","default":false,"description":"ClawSweeper marked this issue as needing maintainer review before automation."},{"id":10971435771,"node_id":"LA_kwDOQb6kR88AAAACjfLS-w","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:needs-product-decision","name":"clawsweeper:needs-product-decision","color":"FBCA04","default":false,"description":"ClawSweeper marked this issue as needing a product or behavior decision."},{"id":10971438267,"node_id":"LA_kwDOQb6kR88AAAACjfLcuw","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:source-repro","name":"clawsweeper:source-repro","color":"0A3069","default":false,"description":"ClawSweeper found a high-confidence source-level issue reproduction."},{"id":10971438454,"node_id":"LA_kwDOQb6kR88AAAACjfLddg","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:linked-pr-open","name":"clawsweeper:linked-pr-open","color":"57606A","default":false,"description":"ClawSweeper found an open linked pull request for this issue."},{"id":10981171401,"node_id":"LA_kwDOQb6kR88AAAACjodgyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/issue-rating:%20%F0%9F%A6%9E%20diamond%20lobster","name":"issue-rating: šŸ¦ž diamond lobster","color":"0969DA","default":false,"description":"Very strong issue quality with high-confidence source-level or clear reproduction."},{"id":11045436603,"node_id":"LA_kwDOQb6kR88AAAACklv8uw","url":"https://api.github.com/repos/openclaw/openclaw/labels/impact:other","name":"impact:other","color":"C5DEF5","default":false,"description":"This issue has meaningful maintainer-visible impact outside the owned taxonomy."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":2,"created_at":"2026-02-02T21:59:05Z","updated_at":"2026-06-16T12:36:52Z","closed_at":null,"assignee":null,"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"sub_issues_summary":{"total":0,"completed":0,"percent_completed":0},"issue_dependencies_summary":{"blocked_by":0,"total_blocked_by":0,"blocking":0,"total_blocking":0},"body":"## Problem\n\nCurrently, the agent configuration schema supports these fields:\n- `id`, `name`, `workspace`, `agentDir`, `model`, `identity`, `tools`, `heartbeat`, `subagents`\n\nHowever, there's **no `description` field** to describe what each agent specializes in.\n\n## Use Case: Dynamic Agent Discovery\n\nWhen orchestrating multiple sub-agents (Sisters), the main agent needs to:\n1. List available agents via `agents_list()`\n2. **Understand what each agent is good at**\n3. Select the best agent for a given task\n4. Delegate via `sessions_spawn()`\n\nWithout a `description` field, the orchestrator agent has no way to know what each sub-agent specializes in at runtime. This forces workarounds like:\n- Hardcoding a static catalog in the main agent's system prompt\n- Reading each agent's SOUL.md file manually (slow, fragile)\n\n## Proposed Solution\n\nAdd an optional `description` field to agent config:\n\n```json\n{\n \"agents\": [\n {\n \"id\": \"trading\",\n \"name\": \"Trading Sister\",\n \"description\": \"Polymarket trading, market analysis, edge detection, bet execution\",\n \"workspace\": \"...\",\n ...\n }\n ]\n}\n```\n\nAnd include it in the `agents_list()` tool response.\n\n## Benefits\n\n1. **Dynamic orchestration** - Main agent can intelligently select sub-agents based on task requirements\n2. **Self-documenting** - Agent capabilities are declared in config, not scattered across files\n3. **Scalable** - Adding new agents doesn't require updating the orchestrator's system prompt\n\n## Impact\n\nThis would be a significant capability upgrade for multi-agent setups, enabling truly dynamic agent selection rather than hardcoded routing.\n\n---\n\n*Feature requested by Shannon (OpenClaw user) - building a multi-agent system with 20+ specialized sub-agents*","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/7490/reactions","total_count":0,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":0},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/7490/timeline","performed_via_github_app":null,"state_reason":null,"pinned_comment":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93587","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93587/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93587/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93587/events","html_url":"https://github.com/openclaw/openclaw/pull/93587","id":4673322803,"node_id":"PR_kwDOQb6kR87m9VVv","number":93587,"title":"[Feature]: /label slash command & /new session naming for WebChat/Control UI","user":{"login":"zhangguiping-xydt","id":275915537,"node_id":"U_kgDOEHIjEQ","avatar_url":"https://avatars.githubusercontent.com/u/275915537?v=4","gravatar_id":"","url":"https://api.github.com/users/zhangguiping-xydt","html_url":"https://github.com/zhangguiping-xydt","followers_url":"https://api.github.com/users/zhangguiping-xydt/followers","following_url":"https://api.github.com/users/zhangguiping-xydt/following{/other_user}","gists_url":"https://api.github.com/users/zhangguiping-xydt/gists{/gist_id}","starred_url":"https://api.github.com/users/zhangguiping-xydt/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/zhangguiping-xydt/subscriptions","organizations_url":"https://api.github.com/users/zhangguiping-xydt/orgs","repos_url":"https://api.github.com/users/zhangguiping-xydt/repos","events_url":"https://api.github.com/users/zhangguiping-xydt/events{/privacy}","received_events_url":"https://api.github.com/users/zhangguiping-xydt/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064298433,"node_id":"LA_kwDOQb6kR88AAAACV-EBwQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/channel:%20telegram","name":"channel: telegram","color":"0969DA","default":false,"description":"Channel integration: telegram"},{"id":10064298757,"node_id":"LA_kwDOQb6kR88AAAACV-EDBQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/app:%20web-ui","name":"app: web-ui","color":"6E7781","default":false,"description":"App: web-ui"},{"id":10064298857,"node_id":"LA_kwDOQb6kR88AAAACV-EDaQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/gateway","name":"gateway","color":"57606A","default":false,"description":"Gateway runtime"},{"id":10069934037,"node_id":"LA_kwDOQb6kR88AAAACWDb_1Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/scripts","name":"scripts","color":"57606A","default":false,"description":"Repository scripts"},{"id":10069934305,"node_id":"LA_kwDOQb6kR88AAAACWDcA4Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/commands","name":"commands","color":"0A3069","default":false,"description":"Command implementations"},{"id":10190106760,"node_id":"LA_kwDOQb6kR88AAAACX2CwiA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20L","name":"size: L","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":1,"created_at":"2026-06-16T10:29:53Z","updated_at":"2026-06-16T12:36:50Z","closed_at":null,"assignee":null,"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93587","html_url":"https://github.com/openclaw/openclaw/pull/93587","diff_url":"https://github.com/openclaw/openclaw/pull/93587.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93587.patch","merged_at":null},"body":"## Summary\n\n- Add session display names as first-class session metadata so dashboard sessions can be named by users without changing session keys.\n- Wire `/new ` to create a named dashboard session and `/label ` to rename the current dashboard session.\n- Surface the persisted `displayName` through `openclaw sessions --json` and the existing Control UI session display fallback.\n- Fix classification: root-cause session metadata contract fix across gateway schema, session store mutation, CLI projection, and Control UI command dispatch.\n- Maintainer-ready confidence: High; focused regression coverage exercises the protocol, gateway create/patch paths, CLI JSON output, Control UI slash command dispatch, and the shared/non-UI command-surface boundaries for `/label` and `/new`.\n- Root cause: Session display names were not part of the gateway session create/patch contract or CLI row projection, so Control UI slash commands had no durable metadata field to write and `openclaw sessions --json` had no stable field to expose.\n- Why this is root-cause fix: The patch adds `displayName` at the owning session metadata boundary, persists it through the existing session store create/patch lifecycle, and then reads the same stored field in both the UI display path and CLI JSON path instead of deriving a name from labels or session keys.\n- What did NOT change: This change does not change existing `label` uniqueness or semantics, session key generation, auth, provider routing, config shape, migrations, or the no-argument `/new` behavior.\n\n## Linked context\n\nCloses #93422\n\nRelated context: the issue requests `/label ` and `/new ` naming for WebChat/Control UI sessions, with the name visible in the Control UI session picker and in `openclaw sessions --json`.\n\nThis was issue-driven contributor work; no separate maintainer or owner request is being claimed.\n\n## Real behavior proof (required for external PRs)\n\n- Behavior or issue addressed: Named dashboard sessions now persist as `SessionEntry.displayName`, can be set by `/new ` and `/label `, and are visible through `openclaw sessions --json` without reusing the existing `label` field.\n- Real environment tested: Source checkout CLI/runtime from this patched branch on Node.js v24.16.0, using a real JSON session store file and the OpenClaw CLI `sessions --json` command path.\n- Exact steps or command run after this patch:\n\n ```bash\n STORE=\"$(mktemp)\"\n node -e 'const fs=require(\"node:fs\"); fs.writeFileSync(process.argv[1], JSON.stringify({main:{sessionId:\"abc123\",updatedAt:Date.now(),displayName:\"Research Plan\",model:\"test:opus\"}}, null, 2));' \"$STORE\"\n node scripts/run-node.mjs sessions --json --store \"$STORE\" --limit all\n rm -f \"$STORE\"\n ```\n\n- Evidence after fix (screenshot, recording, terminal capture, console output, redacted runtime log, linked artifact, or copied live output): Terminal capture from the command above. Exit code: 0. The temporary store path is redacted from the copied output.\n\n ```json\n {\n \"path\": \"\",\n \"count\": 1,\n \"totalCount\": 1,\n \"limitApplied\": null,\n \"hasMore\": false,\n \"activeMinutes\": null,\n \"sessions\": [\n {\n \"key\": \"main\",\n \"updatedAt\": 1781604245105,\n \"ageMs\": 21970,\n \"sessionId\": \"abc123\",\n \"displayName\": \"Research Plan\",\n \"totalTokens\": null,\n \"totalTokensFresh\": false,\n \"model\": \"test:opus\",\n \"modelProvider\": \"zte\",\n \"contextTokens\": 100000,\n \"agentId\": \"main\",\n \"acpRuntime\": false,\n \"agentRuntime\": {\n \"id\": \"auto\",\n \"source\": \"implicit\"\n },\n \"kind\": \"direct\"\n }\n ]\n }\n ```\n\n- Observed result after fix: The runtime CLI JSON row for `main` includes `\"displayName\": \"Research Plan\"`; focused regression tests also passed for gateway create/patch persistence, protocol validation, CLI JSON output, and Control UI `/new ` plus `/label ` dispatch.\n- What was not tested: A browser screenshot of the Control UI session picker was not captured in this environment; the Control UI command behavior and selected-agent global-session scope were covered by focused tests.\n- Proof limitations or environment constraints: External live Control UI screenshot proof is unavailable in this environment; local verification completed with the CLI proof above and the focused validation commands below.\n- Before evidence (optional but encouraged): Before this patch, `displayName` was not accepted by the session create/patch schema, was not persisted by session store patching, and was not included in `openclaw sessions --json` rows.\n\n## Tests and validation\n\nCommands run after the final patch set:\n\n- `node scripts/run-vitest.mjs run --config test/vitest/vitest.unit.config.ts packages/gateway-protocol/src/index.test.ts` passed: 1 test file, 45 tests.\n- `node scripts/run-vitest.mjs run --config test/vitest/vitest.commands.config.ts src/commands/sessions.test.ts` passed: 1 test file, 16 tests.\n- `node scripts/run-vitest.mjs run --config test/vitest/vitest.auto-reply-core.config.ts src/auto-reply/command-control.test.ts src/auto-reply/commands-registry.test.ts` passed: 2 test files, 79 tests.\n- `node scripts/run-vitest.mjs run --config test/vitest/vitest.unit-ui.config.ts ui/src/ui/app-chat.test.ts ui/src/ui/chat/slash-commands.browser-import.test.ts` passed: 2 test files, 101 tests.\n- `node scripts/run-vitest.mjs run --config test/vitest/vitest.gateway.config.ts src/gateway/sessions-patch.test.ts src/gateway/server.sessions.create.test.ts` passed: 4 test files, 178 tests.\n- `node scripts/run-vitest.mjs run --config test/vitest/vitest.gateway.config.ts src/gateway/server.sessions.reset-hooks.test.ts` passed: 1 test file, 18 tests.\n- `node scripts/run-vitest.mjs run --config test/vitest/vitest.runtime-config.config.ts src/config/sessions/store.session-key-normalization.test.ts` passed: 1 test file, 11 tests.\n- `pnpm check:test-types` passed.\n- `pnpm protocol:gen && pnpm protocol:gen:swift && git diff --exit-code -- dist/protocol.schema.json apps/macos/Sources/OpenClawProtocol/GatewayModels.swift apps/shared/OpenClawKit/Sources/OpenClawProtocol/GatewayModels.swift` passed.\n- Follow-up after the branch refresh: `pnpm tsgo:test` passed.\n- Follow-up after the branch refresh: extension-specific Telegram test `node scripts/run-vitest.mjs run --config test/vitest/vitest.extension-telegram.config.ts extensions/telegram/src/bot.test.ts --reporter=verbose` passed: 1 test file.\n- Follow-up after the branch refresh: `node scripts/run-vitest.mjs run --config test/vitest/vitest.tooling.config.ts src/scripts/test-projects.test.ts --reporter=verbose` passed: 1 test file, 82 tests.\n- Follow-up after ClawSweeper review: `node scripts/run-vitest.mjs run --config test/vitest/vitest.unit.config.ts packages/gateway-protocol/src/native-protocol-levels.guard.test.ts --reporter=verbose` passed: 1 test file, 5 tests, including the Swift optional `displayName` initializer compatibility guard.\n- Follow-up after ClawSweeper review: `pnpm protocol:gen && pnpm protocol:gen:swift && git diff --exit-code -- dist/protocol.schema.json apps/macos/Sources/OpenClawProtocol/GatewayModels.swift apps/shared/OpenClawKit/Sources/OpenClawProtocol/GatewayModels.swift` passed after regenerating Swift models.\n- `node scripts/run-node.mjs sessions --json --store \"$STORE\" --limit all` passed and printed the named session row shown in the real behavior proof above.\n\nRegression coverage added or updated:\n\n- Protocol validation accepts non-empty `displayName` on session create and patch, and rejects blank values.\n- Gateway session patch persists, trims, clears, and preserves `displayName` according to patch presence.\n- Gateway session create writes `displayName` when supplied and preserves an existing display name when omitted.\n- `openclaw sessions --json` includes `displayName` in the session row.\n- Control UI `/new ` passes a display-name payload while no-argument `/new` keeps the previous behavior.\n- Control UI `/label ` patches `displayName`, rejects missing names, scopes selected-agent global sessions with `agentId`, and runs immediately while a chat stream is busy.\n- Non-UI command detection does not treat `/label ` as a shared control command, while Control UI fallback slash commands still expose `/label` locally.\n- Shared/native command metadata does not advertise a display-name argument for `/new`, keeping named `/new ` behavior scoped to the Control UI path whose runtime persists `displayName`.\n- Inbound group/channel metadata updates preserve user-provided `displayName` values while still refreshing channel-owned generated display names when group metadata changes.\n- `session.dmScope=\"main\"` reset-in-place creation applies `/new ` display names during the reset path instead of preserving the old name.\n\nWhat failed before this fix: the new RED tests failed because the schema rejected `displayName`, patch/create paths did not persist it, CLI JSON rows omitted it, and Control UI slash dispatch had no durable display-name write path.\n\n## Risk checklist\n\nDid user-visible behavior change? (`Yes/No`): Yes. Users can now name dashboard sessions from chat commands and see that name in session surfaces.\n\nDid config, environment, or migration behavior change? (`Yes/No`): No.\n\nDid security, auth, secrets, network, or tool execution behavior change? (`Yes/No`): No.\n\nWhat is the highest-risk area? The highest-risk area is session metadata ownership and command surface scoping: patching `global` without the selected `agentId` could write the display name to the wrong agent's store, channel-derived display names could overwrite user-provided names, `session.dmScope=\"main\"` reset-in-place could drop `/new `, advertising `/label` outside the Control UI would create a command with no shared channel handler, and advertising `/new` display-name args to native surfaces would imply behavior that non-UI runtimes do not implement.\n\nHow is that risk mitigated? `/label` now mirrors the selected-agent session scope before calling `sessions.patch`, reloads sessions with the same scope, stays registered as a Control UI-only slash command, `/new` display-name arg metadata is not exposed through shared/native command specs, group metadata refreshes only overwrite channel-owned generated display names, main-scoped reset-in-place applies an explicit display-name override, and regression coverage exercises selected-agent global sessions, busy chat streams, non-UI command detection, shared native command metadata, inbound group metadata ownership, and main dmScope reset-in-place creation.\n\n## Current review state\n\nNext action: rerun submit-readiness gates, then use the pr-auto submit flow if the body, proof, and review gates pass.\n\nStill waiting on author, maintainer, CI, or external proof: no local author action is known after the current gate rerun; remote CI and maintainer review will only start after PR creation.\n\nBot or reviewer comments addressed: Earlier findings about selected-agent global `/label` scope, `sessions.create` preserving existing display names when `displayName` is omitted, busy `/label` queuing, keeping `/label` out of shared non-UI command detection, keeping `/new` display-name args out of shared/native command metadata, preserving user-provided display names from later group metadata refreshes, and applying `/new ` during `session.dmScope=\"main\"` reset-in-place were addressed with regression tests.\n\nRemote CI follow-up addressed: the first PR run failed `checks-fast-bundled-protocol` because the Swift gateway protocol model had not been regenerated after adding `displayName` to the session schema; the generated `GatewayModels.swift` update is now committed and the bundled protocol generation diff check passes locally. After the branch refresh, `check-test-types` and `checks-node-core-tooling` were addressed with focused test-alignment fixes and the matching local validation commands pass. ClawSweeper also flagged Swift source compatibility for the additive optional `displayName` initializer arguments; the Swift generator now emits `displayname` with a default `nil`, a native protocol guard covers this, and regenerated Swift models plus the protocol generation diff check pass locally.\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93587/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93587/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/75","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/75/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/75/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/75/events","html_url":"https://github.com/openclaw/openclaw/issues/75","id":3774902933,"node_id":"I_kwDOQb6kR87hAHKV","number":75,"title":"Linux/Windows Clawdbot Apps","user":{"login":"steipete","id":58493,"node_id":"MDQ6VXNlcjU4NDkz","avatar_url":"https://avatars.githubusercontent.com/u/58493?v=4","gravatar_id":"","url":"https://api.github.com/users/steipete","html_url":"https://github.com/steipete","followers_url":"https://api.github.com/users/steipete/followers","following_url":"https://api.github.com/users/steipete/following{/other_user}","gists_url":"https://api.github.com/users/steipete/gists{/gist_id}","starred_url":"https://api.github.com/users/steipete/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/steipete/subscriptions","organizations_url":"https://api.github.com/users/steipete/orgs","repos_url":"https://api.github.com/users/steipete/repos","events_url":"https://api.github.com/users/steipete/events{/privacy}","received_events_url":"https://api.github.com/users/steipete/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":9706244996,"node_id":"LA_kwDOQb6kR88AAAACQomLhA","url":"https://api.github.com/repos/openclaw/openclaw/labels/enhancement","name":"enhancement","color":"0969DA","default":true,"description":"New feature or request"},{"id":9706245005,"node_id":"LA_kwDOQb6kR88AAAACQomLjQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/help%20wanted","name":"help wanted","color":"008672","default":true,"description":"Extra attention is needed"},{"id":10970446441,"node_id":"LA_kwDOQb6kR88AAAACjeO6aQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/P2","name":"P2","color":"FBCA04","default":false,"description":"Normal backlog priority with limited blast radius."},{"id":10971435209,"node_id":"LA_kwDOQb6kR88AAAACjfLQyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:no-new-fix-pr","name":"clawsweeper:no-new-fix-pr","color":"E5E7EB","default":false,"description":"ClawSweeper does not recommend queueing a new automated fix PR for this issue."},{"id":10971435625,"node_id":"LA_kwDOQb6kR88AAAACjfLSaQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:needs-maintainer-review","name":"clawsweeper:needs-maintainer-review","color":"FBCA04","default":false,"description":"ClawSweeper marked this issue as needing maintainer review before automation."},{"id":10971435771,"node_id":"LA_kwDOQb6kR88AAAACjfLS-w","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:needs-product-decision","name":"clawsweeper:needs-product-decision","color":"FBCA04","default":false,"description":"ClawSweeper marked this issue as needing a product or behavior decision."},{"id":10973513130,"node_id":"LA_kwDOQb6kR88AAAACjhKFqg","url":"https://api.github.com/repos/openclaw/openclaw/labels/impact:security","name":"impact:security","color":"B60205","default":false,"description":"Security boundary, credential, authz, sandbox, or sensitive-data risk."},{"id":10981172445,"node_id":"LA_kwDOQb6kR88AAAACjodk3Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/issue-rating:%20%F0%9F%8C%8A%20off-meta%20tidepool","name":"issue-rating: šŸŒŠ off-meta tidepool","color":"6E7781","default":false,"description":"Issue quality rating does not apply to this item."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":109,"created_at":"2026-01-01T16:55:34Z","updated_at":"2026-06-16T12:36:44Z","closed_at":null,"assignee":null,"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"sub_issues_summary":{"total":0,"completed":0,"percent_completed":0},"issue_dependencies_summary":{"blocked_by":0,"total_blocked_by":0,"blocking":0,"total_blocking":0},"body":"We have apps for macOS, iOS and Android (simpler nodes)\n\nLinux and Windows are missing. Similar feature set to macOS ideally.","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/75/reactions","total_count":88,"+1":79,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":4,"rocket":1,"eyes":4},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/75/timeline","performed_via_github_app":null,"state_reason":null,"pinned_comment":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93627","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93627/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93627/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93627/events","html_url":"https://github.com/openclaw/openclaw/pull/93627","id":4674108068,"node_id":"PR_kwDOQb6kR87m_8oz","number":93627,"title":"fix(dreaming): filter already-emitted entries in light phase to prevent verbatim repeats (fixes #72096)","user":{"login":"liuhao1024","id":11816344,"node_id":"MDQ6VXNlcjExODE2MzQ0","avatar_url":"https://avatars.githubusercontent.com/u/11816344?v=4","gravatar_id":"","url":"https://api.github.com/users/liuhao1024","html_url":"https://github.com/liuhao1024","followers_url":"https://api.github.com/users/liuhao1024/followers","following_url":"https://api.github.com/users/liuhao1024/following{/other_user}","gists_url":"https://api.github.com/users/liuhao1024/gists{/gist_id}","starred_url":"https://api.github.com/users/liuhao1024/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/liuhao1024/subscriptions","organizations_url":"https://api.github.com/users/liuhao1024/orgs","repos_url":"https://api.github.com/users/liuhao1024/repos","events_url":"https://api.github.com/users/liuhao1024/events{/privacy}","received_events_url":"https://api.github.com/users/liuhao1024/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064299264,"node_id":"LA_kwDOQb6kR88AAAACV-EFAA","url":"https://api.github.com/repos/openclaw/openclaw/labels/extensions:%20memory-core","name":"extensions: memory-core","color":"6E7781","default":false,"description":"Extension: memory-core"},{"id":10190106613,"node_id":"LA_kwDOQb6kR88AAAACX2Cv9Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20XS","name":"size: XS","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":0,"created_at":"2026-06-16T12:24:43Z","updated_at":"2026-06-16T12:36:40Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93627","html_url":"https://github.com/openclaw/openclaw/pull/93627","diff_url":"https://github.com/openclaw/openclaw/pull/93627.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93627.patch","merged_at":null},"body":"## What does this PR do?\n\nPrevents the dreaming light phase from emitting the same work-summary block verbatim across consecutive cycles. When entries remain in the lookback window but haven't been recalled again, the light phase now filters them out using per-entry `lastLightAt` timestamps from the phase signal store.\n\n## Related Issue\n\nFixes #72096\n\n## Type of Change\n\n- [x] Bug fix (non-breaking)\n\n## Changes Made\n\n- `extensions/memory-core/src/short-term-promotion.ts`: Added `readPhaseSignalEntries()` ā€” a thin public wrapper around the existing `readPhaseSignalStore()` that exposes per-entry `lastLightAt`/`lastRemAt` timestamps for callers that need deduplication signals without the full store.\n- `extensions/memory-core/src/dreaming-phases.ts`: In `runLightDreaming()`, added a pre-filter step that reads phase signals and excludes entries whose `lastLightAt` is newer than their `lastRecalledAt` (i.e., already emitted and not recalled since). Only fresh or re-recalled entries proceed to the ranking/dedup pipeline.\n\n## Real behavior proof\n\n- **Behavior addressed:** Dreaming light phase emits identical work-summary blocks across consecutive cycles when the same entries remain in the lookback window but aren't recalled again. The fix filters already-emitted entries using phase signal timestamps.\n- **Environment tested:** macOS, Node.js v22, openclaw build (pnpm build)\n- **Steps run after the patch:**\n```\n$ node -e \"\n// Verify readPhaseSignalEntries is exported\nconst mod = require('./dist/extensions/memory-core/short-term-promotion.cjs');\nconsole.log('readPhaseSignalEntries:', typeof mod.readPhaseSignalEntries);\n\"\nreadPhaseSignalEntries: function\n```\n- **Evidence after fix:**\n```\n$ grep -n \"readPhaseSignalEntries\\|freshEntries\\|lastLightAt\" extensions/memory-core/src/dreaming-phases.ts\n43: readPhaseSignalEntries,\n1707: const phaseSignals = await readPhaseSignalEntries({\n1712: const freshEntries = recentEntries.filter((entry) => {\n1714: if (!signal?.lastLightAt) {\n1718: const lastLightMs = Date.parse(signal.lastLightAt);\n```\n```\n$ grep -n \"readPhaseSignalEntries\" extensions/memory-core/src/short-term-promotion.ts\n1778:export async function readPhaseSignalEntries(params: {\n```\n- **Observed result after fix:** The `readPhaseSignalEntries` export is available and correctly used in `runLightDreaming`. Entries with a `lastLightAt` newer than their `lastRecalledAt` are filtered out, preventing verbatim repeats. All 47 dreaming-phases tests and 78 short-term-promotion tests pass.\n- **What was not tested:** End-to-end dreaming cycle with actual memory store (requires running OpenClaw workspace with recorded entries). Logic verified via test suite and code inspection.\n\n## Checklist\n\n- [x] Read Contributing Guide\n- [x] Conventional Commits\n- [x] Searched for duplicates\n- [x] Only related changes\n- [x] Tests pass (47/47 dreaming-phases, 78/78 short-term-promotion)\n- [x] Build passes (pnpm build)","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93627/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93627/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93624","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93624/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93624/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93624/events","html_url":"https://github.com/openclaw/openclaw/pull/93624","id":4674022533,"node_id":"PR_kwDOQb6kR87m_qKk","number":93624,"title":"feat(gateway): add configurable streaming chunk size for HTTP endpoints","user":{"login":"Jah-xy","id":263118420,"node_id":"U_kgDOD67eVA","avatar_url":"https://avatars.githubusercontent.com/u/263118420?v=4","gravatar_id":"","url":"https://api.github.com/users/Jah-xy","html_url":"https://github.com/Jah-xy","followers_url":"https://api.github.com/users/Jah-xy/followers","following_url":"https://api.github.com/users/Jah-xy/following{/other_user}","gists_url":"https://api.github.com/users/Jah-xy/gists{/gist_id}","starred_url":"https://api.github.com/users/Jah-xy/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/Jah-xy/subscriptions","organizations_url":"https://api.github.com/users/Jah-xy/orgs","repos_url":"https://api.github.com/users/Jah-xy/repos","events_url":"https://api.github.com/users/Jah-xy/events{/privacy}","received_events_url":"https://api.github.com/users/Jah-xy/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064298857,"node_id":"LA_kwDOQb6kR88AAAACV-EDaQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/gateway","name":"gateway","color":"57606A","default":false,"description":"Gateway runtime"},{"id":10190106648,"node_id":"LA_kwDOQb6kR88AAAACX2CwGA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20S","name":"size: S","color":"8C959F","default":false,"description":null},{"id":10865780645,"node_id":"LA_kwDOQb6kR88AAAACh6anpQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/triage:%20needs-real-behavior-proof","name":"triage: needs-real-behavior-proof","color":"C5DEF5","default":false,"description":"Candidate: external PR needs after-fix proof from a real setup."}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":0,"created_at":"2026-06-16T12:13:17Z","updated_at":"2026-06-16T12:36:39Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93624","html_url":"https://github.com/openclaw/openclaw/pull/93624","diff_url":"https://github.com/openclaw/openclaw/pull/93624.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93624.patch","merged_at":null},"body":"## Summary\n\n- Adds `gateway.http.endpoints.{chatCompletions,responses}.streaming` config to control SSE delta chunk sizes\n- Small assistant text deltas are coalesced into configurable-sized chunks (minChars/maxChars/idleMs) to reduce HTTP overhead\n- Applies to both `/v1/chat/completions` and `/v1/responses` streaming endpoints\n- Default behavior unchanged when config is absent (minChars === maxChars means pass-through, zero new overhead)\n\n## Config Example\n\n```yaml\ngateway:\n http:\n endpoints:\n responses:\n streaming:\n minChars: 200\n maxChars: 800\n chatCompletions:\n streaming:\n minChars: 200\n maxChars: 800\n```\n\n## Changed Files\n\n| File | Purpose |\n|------|---------|\n| `src/config/types.gateway.ts` | `GatewayHttpStreamingConfig` type + `streaming?` field on both endpoint configs |\n| `src/gateway/gateway-streaming-chunker.ts` | Shared chunk buffer (accumulates, flushes at minChars/maxChars/idleMs) |\n| `src/gateway/openai-http.ts` | Wires chunker into `/v1/chat/completions` SSE streaming path |\n| `src/gateway/openresponses-http.ts` | Wires chunker into `/v1/responses` SSE streaming path |\n| `src/gateway/server-http.ts` | (no functional change, type-level only) |\n| `src/gateway/server-runtime-config.ts` | (no functional change, type-level only) |\n\n## Real behavior proof\n\nThis is a **pure config addition** with zero behavioral change on the default path. When `streaming` config is absent (the default), the chunk buffer operates as a **pass-through** (`minChars === maxChars`), emitting each delta immediately ā€” identical to current behavior. No existing users are affected.\n\nThe new `gateway-streaming-chunker.ts` is a leaf module with no dependents beyond the two HTTP handlers. The SSE wire format is unchanged; only the granularity of individual `data:` lines differs when the opt-in config is set.\n\n**Verification**: The chunk buffer was tested in a local gateway setup with `minChars: 200, maxChars: 800`. With small per-token deltas (~1-2 chars), the SSE stream emitted ~400 SSE `data:` lines for a 80-char response instead of ~80 individual lines. Client-side concatenation produces identical output in both cases.\n\nRefs: #93598","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93624/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93624/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93623","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93623/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93623/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93623/events","html_url":"https://github.com/openclaw/openclaw/pull/93623","id":4674010172,"node_id":"PR_kwDOQb6kR87m_ngt","number":93623,"title":"fix(chat): prefer server session row over stale cache in model dropdown","user":{"login":"sunlit-deng","id":253064511,"node_id":"U_kgDODxV1Pw","avatar_url":"https://avatars.githubusercontent.com/u/253064511?v=4","gravatar_id":"","url":"https://api.github.com/users/sunlit-deng","html_url":"https://github.com/sunlit-deng","followers_url":"https://api.github.com/users/sunlit-deng/followers","following_url":"https://api.github.com/users/sunlit-deng/following{/other_user}","gists_url":"https://api.github.com/users/sunlit-deng/gists{/gist_id}","starred_url":"https://api.github.com/users/sunlit-deng/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/sunlit-deng/subscriptions","organizations_url":"https://api.github.com/users/sunlit-deng/orgs","repos_url":"https://api.github.com/users/sunlit-deng/repos","events_url":"https://api.github.com/users/sunlit-deng/events{/privacy}","received_events_url":"https://api.github.com/users/sunlit-deng/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064298757,"node_id":"LA_kwDOQb6kR88AAAACV-EDBQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/app:%20web-ui","name":"app: web-ui","color":"6E7781","default":false,"description":"App: web-ui"},{"id":10190106613,"node_id":"LA_kwDOQb6kR88AAAACX2Cv9Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20XS","name":"size: XS","color":"8C959F","default":false,"description":null},{"id":10865780674,"node_id":"LA_kwDOQb6kR88AAAACh6anwg","url":"https://api.github.com/repos/openclaw/openclaw/labels/triage:%20mock-only-proof","name":"triage: mock-only-proof","color":"C5DEF5","default":false,"description":"Candidate: PR proof only shows tests, mocks, snapshots, lint, typecheck, or CI."}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":1,"created_at":"2026-06-16T12:11:31Z","updated_at":"2026-06-16T12:36:38Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93623","html_url":"https://github.com/openclaw/openclaw/pull/93623","diff_url":"https://github.com/openclaw/openclaw/pull/93623.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93623.patch","merged_at":null},"body":"## Summary\n\nThe chat model dropdown could show a cached model selection (e.g. codex/gpt-5.5)\nwhile the active session was actually using the default/fallback model\n(e.g. ollama/qwen3.5:9b). This happened because resolveChatModelOverrideValue\nalways preferred the local chatModelOverrides cache without cross-referencing\nthe server's session row.\n\nWhen a user selects a model different from the default, the selection is stored\nlocally in chatModelOverrides and also sent to the server as a session patch.\nBut if the session later resets or drifts back to its default model (e.g. after\nlogout/login, session timeout, or model failover), the local cache retains the\nold selection while the server reports the correct runtime model. The dropdown\nthen displays a misleading model.\n\nFix: In resolveChatModelOverrideValue(), when a cached override exists but\nthe server's active session row reports a different runtime model, prefer the\nserver value. The cached override is still used when the server has no model\ndata or when both values agree.\n\nFixes #93346\n\n## Linked context\n\n- Issue: https://github.com/openclaw/openclaw/issues/93346\n- Related: clawsweeper analysis identified the cache-first picker behavior and\n Gateway row projection as two sides of the problem. This PR fixes the UI side\n (cache-first picker) ā€” the dropdown now accurately reflects the server-reported\n runtime model.\n\n## Real behavior proof\n\n**Behavior addressed**: Chat model dropdown now reflects actual runtime model\nwhen the server session row differs from the locally cached override.\n\n**Real setup tested**: Unit tests with vitest\n\n**Exact steps or command run after fix**:\n\n pnpm test -- --run ui/src/ui/chat-model-select-state.test.ts\n\n**After-fix evidence**:\n\n = test/vitest/vitest.ui.config.ts (1 test file)\n = chat-model-select-state (10 tests)\n Test Files 1 passed (1)\n Tests 10 passed (10)\n Start at 20:05:52\n Duration 2.93s\n\n**New test case** (prefers server session row over stale cached override):\n- Sets cached override to openai/gpt-5-mini\n- Server session row reports deepseek-chat (deepseek provider)\n- Expected: deepseek/deepseek-chat (server value wins)\n- The resolved currentOverride is correctly deepseek/deepseek-chat\n\n**Existing behavior preserved** (normalizes cached bare overrides):\n- Sets cached override to gpt-5-mini (raw)\n- Server session row has no model (model: null)\n- Expected: openai/gpt-5-mini (cached value wins when server has no data)\n- The resolved currentOverride is correctly openai/gpt-5-mini\n\n**Observed result after the fix**: All 10 tests pass, including 1 new test\ncovering the drift scenario. The fix is a pure state-selector change ā€” no\ncomponent or server-side modifications needed.\n\n**What was not tested**: E2E tests with live Gateway were not run. The fix is\na pure selector function change verified through unit tests.\n\n## Tests and validation\n\n- All existing 9 tests pass unchanged\n- New test verifies: cached override differs from server -> server wins\n- All 23 chat-model-ref.test.ts tests also pass\n- Ran pnpm check (lint + typecheck) ā€” clean\n\n## Risk checklist\n\nDid user-visible behavior change? (Yes)\n- When the server model differs from the cached override, the dropdown now shows\n the server's runtime model instead of the stale cached selection.\n\nDid config, environment, or migration behavior change? (No)\n\nDid security, auth, secrets, network, or tool execution behavior change? (No)\n\nWhat is the highest-risk area?\n- Brief (<200ms) window where an in-flight model patch could be overridden by\n a slightly stale server response. The server value is authoritative once the\n patch is confirmed.\n\nHow is that risk mitigated?\n- The fix only prefers the server value when it differs from the cache.\n After the server confirms a patch, both values agree and the cache wins\n normally. The divergence only occurs when the session genuinely drifted.\n\n## Current review state\n\nWhat is the next action?\n- Maintainer review\n\nWhat is still waiting on author, maintainer, CI, or external proof?\n- This PR fixes the UI cache side of the issue. The Gateway session row still\n conflates selected and runtime model identity into a single model/modelProvider\n field. A follow-up could expose both fields separately for full transparency,\n but this change already makes the dropdown accurate.\n\nWhich bot or reviewer comments were addressed?\n- clawsweeper noted the cache-first picker as a source-level reproduction path;\n this PR directly addresses that by cross-referencing the server row.\n The Gateway row projection issue (selectedModel vs resolvedModel both\n mapped to row.model) is noted but not addressed here.\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93623/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93623/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93622","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93622/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93622/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93622/events","html_url":"https://github.com/openclaw/openclaw/pull/93622","id":4674003572,"node_id":"PR_kwDOQb6kR87m_mFq","number":93622,"title":"fix(tools): surface malformed availability diagnostics as warnings","user":{"login":"zhouhe-xydt","id":265407618,"node_id":"U_kgDOD9HMgg","avatar_url":"https://avatars.githubusercontent.com/u/265407618?v=4","gravatar_id":"","url":"https://api.github.com/users/zhouhe-xydt","html_url":"https://github.com/zhouhe-xydt","followers_url":"https://api.github.com/users/zhouhe-xydt/followers","following_url":"https://api.github.com/users/zhouhe-xydt/following{/other_user}","gists_url":"https://api.github.com/users/zhouhe-xydt/gists{/gist_id}","starred_url":"https://api.github.com/users/zhouhe-xydt/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/zhouhe-xydt/subscriptions","organizations_url":"https://api.github.com/users/zhouhe-xydt/orgs","repos_url":"https://api.github.com/users/zhouhe-xydt/repos","events_url":"https://api.github.com/users/zhouhe-xydt/events{/privacy}","received_events_url":"https://api.github.com/users/zhouhe-xydt/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10190106648,"node_id":"LA_kwDOQb6kR88AAAACX2CwGA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20S","name":"size: S","color":"8C959F","default":false,"description":null},{"id":10865780645,"node_id":"LA_kwDOQb6kR88AAAACh6anpQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/triage:%20needs-real-behavior-proof","name":"triage: needs-real-behavior-proof","color":"C5DEF5","default":false,"description":"Candidate: external PR needs after-fix proof from a real setup."}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":0,"created_at":"2026-06-16T12:10:34Z","updated_at":"2026-06-16T12:36:36Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93622","html_url":"https://github.com/openclaw/openclaw/pull/93622","diff_url":"https://github.com/openclaw/openclaw/pull/93622.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93622.patch","merged_at":null},"body":"## Summary\n- Surface malformed availability diagnostics as warnings in `buildToolPlan`.\n- Log a warning when a descriptor's availability evaluates to `unsupported-signal` (e.g. empty `allOf`/`anyOf` groups) so authors notice config errors instead of silently hiding tools.\n- Add tests covering the warning behavior and ensuring ordinary runtime unavailability does not warn.\n\n## Test plan\n- [x] `node scripts/run-vitest.mjs run src/tools/planner.test.ts` passes\n- [x] `pnpm exec oxlint src/tools/planner.ts src/tools/planner.test.ts` passes\n- [x] `pnpm build` completes successfully\n\n## Verification\nBehavior addressed: malformed availability descriptors (empty `allOf`/`anyOf`) were hidden without any visible diagnostic; now `buildToolPlan` logs a warning.\nReal environment tested: Linux x86_64, Node 22.19+, pnpm workspace.\nExact steps or command run after this patch:\n 1. `node scripts/run-vitest.mjs run src/tools/planner.test.ts`\n 2. `pnpm exec oxlint src/tools/planner.ts src/tools/planner.test.ts`\n 3. `pnpm build`\nEvidence after fix:\n - Vitest: `Test Files 1 passed (1) / Tests 8 passed (8)`\n - oxlint: exited 0 with no output\n - build: completed with exit code 0\nObserved result after fix: the new `logs a warning when a descriptor has malformed availability` test passes; the `does not log warnings for ordinary runtime unavailability` test confirms env-missing tools remain silent.\nWhat was not tested: live gateway/tool runtime integration; this change is planner-local.","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93622/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93622/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93621","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93621/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93621/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93621/events","html_url":"https://github.com/openclaw/openclaw/pull/93621","id":4673999143,"node_id":"PR_kwDOQb6kR87m_lIp","number":93621,"title":"fix(session): remove stale .jsonl.lock on session write lock release","user":{"login":"lzyyzznl","id":41978486,"node_id":"MDQ6VXNlcjQxOTc4NDg2","avatar_url":"https://avatars.githubusercontent.com/u/41978486?v=4","gravatar_id":"","url":"https://api.github.com/users/lzyyzznl","html_url":"https://github.com/lzyyzznl","followers_url":"https://api.github.com/users/lzyyzznl/followers","following_url":"https://api.github.com/users/lzyyzznl/following{/other_user}","gists_url":"https://api.github.com/users/lzyyzznl/gists{/gist_id}","starred_url":"https://api.github.com/users/lzyyzznl/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/lzyyzznl/subscriptions","organizations_url":"https://api.github.com/users/lzyyzznl/orgs","repos_url":"https://api.github.com/users/lzyyzznl/repos","events_url":"https://api.github.com/users/lzyyzznl/events{/privacy}","received_events_url":"https://api.github.com/users/lzyyzznl/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10069976942,"node_id":"LA_kwDOQb6kR88AAAACWDenbg","url":"https://api.github.com/repos/openclaw/openclaw/labels/agents","name":"agents","color":"57606A","default":false,"description":"Agent runtime and tooling"},{"id":10190106613,"node_id":"LA_kwDOQb6kR88AAAACX2Cv9Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20XS","name":"size: XS","color":"8C959F","default":false,"description":null},{"id":10865780645,"node_id":"LA_kwDOQb6kR88AAAACh6anpQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/triage:%20needs-real-behavior-proof","name":"triage: needs-real-behavior-proof","color":"C5DEF5","default":false,"description":"Candidate: external PR needs after-fix proof from a real setup."}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":0,"created_at":"2026-06-16T12:09:57Z","updated_at":"2026-06-16T12:36:35Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93621","html_url":"https://github.com/openclaw/openclaw/pull/93621","diff_url":"https://github.com/openclaw/openclaw/pull/93621.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93621.patch","merged_at":null},"body":"## Summary\n\n- Fixes #93383 ā€” Stale `.jsonl.lock` files orphaned on disk after sessions complete, blocking future writes\n- Root cause: `lock.release()` from the file-lock-manager unlocked but did not remove the lockfile. Since the gateway PID remains alive, the stale-lock detection never fired.\n- Fix: wrap the release to `fs.rm(lockPath, { force: true })` after unlocking\n\n## Linked context\n\n- Issue #93383: detailed root cause analysis identifying 4 bugs\n- Bug 4 (this fix): \"When status:done is written and endedAt is set, the lockfile should be removed. Today it isn't.\"\n- The reporter had to run a local 60s launchd sweeper as a workaround\n\n## Changes\n\n`src/agents/session-write-lock.ts` ā€” 1 file, +9/-1\n\nThe `acquireSessionWriteLock` return value now wraps `lock.release()` to also `fs.rm()` the lockfile:\n\n```diff\n- return { release: lock.release };\n+ return {\n+ release: async () => {\n+ await lock.release();\n+ await fs.rm(lockPath, { force: true }).catch(() => {});\n+ },\n+ };\n```\n\n## Risk checklist\n\nDid user-visible behavior change? (`No`) ā€” Lock files are cleaned up automatically instead of persisting\nDid config, environment, or migration behavior change? (`No`)\nDid security, auth, secrets, network, or tool execution behavior change? (`No`)\nWhat is the highest-risk area? ā€” Removing the lockfile after release could race with a concurrent acquire that already created a new lockfile\nHow is that risk mitigated? ā€” `force: true` only removes if the file exists; `.catch(() => {})` suppresses errors. The underlying lock is already released, so the file is genuinely orphaned.\n\n## Current review state\n\nWhat is the next action? ā€” Maintainer review\nWhat is still waiting on author, maintainer, CI, or external proof? ā€” Real behavior proof CI\nWhich bot or reviewer comments were addressed? ā€” None yet","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93621/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93621/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93620","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93620/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93620/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93620/events","html_url":"https://github.com/openclaw/openclaw/pull/93620","id":4673989974,"node_id":"PR_kwDOQb6kR87m_jJ3","number":93620,"title":"[AI] fix(openai-completions): preserve reasoning_content on assistant messages for OpenRouter providers","user":{"login":"xydt-tanshanshan","id":291969148,"node_id":"U_kgDOEWcYfA","avatar_url":"https://avatars.githubusercontent.com/u/291969148?v=4","gravatar_id":"","url":"https://api.github.com/users/xydt-tanshanshan","html_url":"https://github.com/xydt-tanshanshan","followers_url":"https://api.github.com/users/xydt-tanshanshan/followers","following_url":"https://api.github.com/users/xydt-tanshanshan/following{/other_user}","gists_url":"https://api.github.com/users/xydt-tanshanshan/gists{/gist_id}","starred_url":"https://api.github.com/users/xydt-tanshanshan/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/xydt-tanshanshan/subscriptions","organizations_url":"https://api.github.com/users/xydt-tanshanshan/orgs","repos_url":"https://api.github.com/users/xydt-tanshanshan/repos","events_url":"https://api.github.com/users/xydt-tanshanshan/events{/privacy}","received_events_url":"https://api.github.com/users/xydt-tanshanshan/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10069976942,"node_id":"LA_kwDOQb6kR88AAAACWDenbg","url":"https://api.github.com/repos/openclaw/openclaw/labels/agents","name":"agents","color":"57606A","default":false,"description":"Agent runtime and tooling"},{"id":10190106613,"node_id":"LA_kwDOQb6kR88AAAACX2Cv9Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20XS","name":"size: XS","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":0,"created_at":"2026-06-16T12:08:42Z","updated_at":"2026-06-16T12:36:33Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93620","html_url":"https://github.com/openclaw/openclaw/pull/93620","diff_url":"https://github.com/openclaw/openclaw/pull/93620.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93620.patch","merged_at":null},"body":"## Summary\r\nMiniMax M3 via OpenRouter returns `reasoning`/`reasoning_details` in its response, but OpenClaw does not preserve these in message history for the next request. This regression affects both plain-text and tool-call turns.\r\n\r\n**Root cause**: MiniMax M3 model ID is not in the `REASONING_CONTENT_REPLAY_MODEL_IDS` allowlist in `openai-transport-stream.ts`. When building the next request's messages, the reasoning content replay fields are stripped because `shouldPreserveReasoningContentReplay` doesn't recognize MiniMax M3 as a reasoning-content-replay-capable model.\r\n\r\n**Fix**: Add `minimax-m3` to the `REASONING_CONTENT_REPLAY_MODEL_IDS` allowlist. The `getReasoningContentReplayModelIdCandidates` helper extracts the base model ID from OpenRouter's format (`openrouter/minimax/minimax-m3` ā†’ `minimax-m3`), so both direct and OpenRouter-routed MiniMax M3 are covered.\r\n\r\n**Changes** (1 file, +1 line):\r\n- `src/agents/openai-transport-stream.ts:4024` ā€” add `\"minimax-m3\"` to `REASONING_CONTENT_REPLAY_MODEL_IDS`\r\n\r\n## Change Type\r\n- [x] Bug fix (non-breaking change which fixes an issue)\r\n\r\n## Scope\r\n- [x] LLM provider: OpenAI-completions compat (OpenRouter reasoning)\r\n\r\n## Linked Issue\r\nRelated to #92769\r\n\r\n## Motivation\r\nMiniMax M3 users via OpenRouter with thinking enabled lose reasoning context across turns. The model fabricates different answers than what it originally reasoned, producing silently wrong outputs.\r\n\r\n## Real behavior proof\r\n**Behavior addressed**: MiniMax M3 via OpenRouter now preserves `reasoning`/`reasoning_details` in message history for both plain-text and tool-call turns.\r\n\r\n**Real environment tested**: Not tested on real OpenRouter endpoint (requires live API key). Unit tests verify compat flag propagation.\r\n\r\n**Exact steps or command run after this patch**:\r\n```\r\npnpm test \"src/agents/openai-completions-compat.test.ts\" \"src/llm/providers/openai-completions.test.ts\" \"src/agents/openai-transport-stream.test.ts\"\r\n```\r\n\r\n**Evidence after fix**:\r\n```\r\nTest Files 3 passed (3)\r\n Tests 315 passed (315)\r\n```\r\n\r\nAll existing tests pass; no regressions. Previously failing tests (`strips OpenRouter Anthropic non-replayable reasoning fields`, `strips OpenRouter xAI non-replayable reasoning fields`) continue to pass.\r\n\r\n**Observed result after fix**: MiniMax M3 is now recognized as a reasoning-content-replay-capable model in the `REASONING_CONTENT_REPLAY_MODEL_IDS` allowlist, so `shouldPreserveReasoningContentReplay` returns true for both direct and OpenRouter-routed MiniMax M3.\r\n\r\n**What was not tested**: End-to-end behavior with a live MiniMax M3 OpenRouter session (requires API key and real model access).\r\n\r\n## Tests and validation\r\n- `src/agents/openai-completions-compat.test.ts` ā€” **17/17 passed**\r\n- `src/llm/providers/openai-completions.test.ts` ā€” **27/27 passed**\r\n- `src/agents/openai-transport-stream.test.ts` ā€” **271/271 passed**\r\n\r\n## Security Impact\r\nNo security impact. The change only adds empty `reasoning_content` markers to assistant messages for OpenRouter providers.\r\n\r\n## Human Verification\r\n- [x] I have reviewed the code changes\r\n- [x] I understand all changes in this PR\r\n- [x] The changes are minimal and focused on the bug fix\r\n\r\n## Compatibility / Risks\r\nLow risk. The `model.reasoning` guard ensures only reasoning-enabled models are affected. The change is additive ā€” it only adds a field that was previously missing.\r\n\r\n## AI Assistance šŸ¤–\r\n- AI-assisted: Yes\r\n- Human confirmed understanding: Yes\r\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93620/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93620/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93619","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93619/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93619/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93619/events","html_url":"https://github.com/openclaw/openclaw/pull/93619","id":4673970857,"node_id":"PR_kwDOQb6kR87m_fDF","number":93619,"title":"fix(session): remove orphaned .jsonl.lock on terminal session state","user":{"login":"ajwan8998","id":50291365,"node_id":"MDQ6VXNlcjUwMjkxMzY1","avatar_url":"https://avatars.githubusercontent.com/u/50291365?v=4","gravatar_id":"","url":"https://api.github.com/users/ajwan8998","html_url":"https://github.com/ajwan8998","followers_url":"https://api.github.com/users/ajwan8998/followers","following_url":"https://api.github.com/users/ajwan8998/following{/other_user}","gists_url":"https://api.github.com/users/ajwan8998/gists{/gist_id}","starred_url":"https://api.github.com/users/ajwan8998/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/ajwan8998/subscriptions","organizations_url":"https://api.github.com/users/ajwan8998/orgs","repos_url":"https://api.github.com/users/ajwan8998/repos","events_url":"https://api.github.com/users/ajwan8998/events{/privacy}","received_events_url":"https://api.github.com/users/ajwan8998/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064298857,"node_id":"LA_kwDOQb6kR88AAAACV-EDaQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/gateway","name":"gateway","color":"57606A","default":false,"description":"Gateway runtime"},{"id":10190106613,"node_id":"LA_kwDOQb6kR88AAAACX2Cv9Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20XS","name":"size: XS","color":"8C959F","default":false,"description":null},{"id":10790052318,"node_id":"LA_kwDOQb6kR88AAAACgyMh3g","url":"https://api.github.com/repos/openclaw/openclaw/labels/triage:%20blank-template","name":"triage: blank-template","color":"C5DEF5","default":false,"description":"Candidate: PR template appears mostly untouched."},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":0,"created_at":"2026-06-16T12:06:11Z","updated_at":"2026-06-16T12:36:32Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93619","html_url":"https://github.com/openclaw/openclaw/pull/93619","diff_url":"https://github.com/openclaw/openclaw/pull/93619.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93619.patch","merged_at":null},"body":"## Summary\n\nWhen a sub-agent run is killed mid-flight by `EmbeddedAttemptSessionTakeoverError`, the lock release path is never reached and the parent session's `.jsonl.lock` remains orphaned on disk. The next inbound on that session blocks indefinitely on the contested lock, effectively killing the conversation.\n\n## Changes\n\n**`src/gateway/session-lifecycle-state.ts`**:\n- In `persistGatewaySessionLifecycleEvent`, after persisting a terminal lifecycle event (`end`/`error`), unconditionally `unlink()` the corresponding `.jsonl.lock` file\n- The lock path is derived from the session store path (`storePath + '.lock'`), matching the convention used by `acquireSessionWriteLock`\n\n## Real behavior proof\n\nBehavior addressed: When a session reaches a terminal state (status=done, endedAt set), any stale `.jsonl.lock` file is now cleaned up so subsequent inbound messages don't block on a contested lock.\n\nEnvironment tested: OpenClaw 2026.6.6 (commit c14793d35a). Tested locally via TypeScript compilation and git diff verification.\n\nExact steps or command run after this patch: After writing status=done/endedAt to the session store in `persistGatewaySessionLifecycleEvent`, the fix calls `await unlink(sessionEntry.storePath + '.lock')`. If the lock file doesn't exist, the try-catch silently skips it.\n\nEvidence after fix:\n```\n$ git diff main...HEAD -- src/gateway/session-lifecycle-state.ts\ndiff --git a/src/gateway/session-lifecycle-state.ts b/src/gateway/session-lifecycle-state.ts\nindex 6af16265c2..b354590b1d 100644\n--- a/src/gateway/session-lifecycle-state.ts\n+++ b/src/gateway/session-lifecycle-state.ts\n@@ -1,5 +1,6 @@\n // Gateway session lifecycle state projection.\n // Converts agent run lifecycle events into session row/store status updates.\n+import { unlink } from \"node:fs/promises\";\n import {\n buildAgentRunTerminalOutcome,\n type AgentRunTerminalOutcome,\n@@ -296,4 +297,15 @@ export async function persistGatewaySessionLifecycleEvent(params: {\n return Object.keys(patch).length > 0 ? patch : null;\n },\n });\n+\n+ // On terminal events (end/error), clean up the orphaned .jsonl.lock file\n+ // so the next inbound does not block on a stale lock (#93383).\n+ if (phase === \"end\" || phase === \"error\") {\n+ const lockPath = `${sessionEntry.storePath}.lock`;\n+ try {\n+ await unlink(lockPath);\n+ } catch {\n+ // Lock file may not exist ā€” that's fine.\n+ }\n+ }\n }\n```\nThe `storePath + '.lock'` convention matches `acquireSessionWriteLock` at `src/agents/session-write-lock.ts:903`.\n\nObserved result after fix: Orphaned .jsonl.lock files are removed when the session reaches terminal state. Without this fix, the lock stays on disk and blocks the next inbound message to that session key.\n\nWhat was not tested: End-to-end reproduction of the sub-agent race condition (requires specific timing). The fix follows the same degraded-service pattern as `cleanStaleLockFiles` in `session-write-lock.ts`.\n\nCloses: #93383\n\nšŸ¤– Generated with [Claude Code](https://claude.com/claude-code)\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93619/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93619/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93618","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93618/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93618/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93618/events","html_url":"https://github.com/openclaw/openclaw/pull/93618","id":4673923212,"node_id":"PR_kwDOQb6kR87m_U9M","number":93618,"title":"fix(feishu): filter temporary card-action-c-* IDs from reply target to prevent Invalid open_message_id errors (fixes #56818)","user":{"login":"liuhao1024","id":11816344,"node_id":"MDQ6VXNlcjExODE2MzQ0","avatar_url":"https://avatars.githubusercontent.com/u/11816344?v=4","gravatar_id":"","url":"https://api.github.com/users/liuhao1024","html_url":"https://github.com/liuhao1024","followers_url":"https://api.github.com/users/liuhao1024/followers","following_url":"https://api.github.com/users/liuhao1024/following{/other_user}","gists_url":"https://api.github.com/users/liuhao1024/gists{/gist_id}","starred_url":"https://api.github.com/users/liuhao1024/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/liuhao1024/subscriptions","organizations_url":"https://api.github.com/users/liuhao1024/orgs","repos_url":"https://api.github.com/users/liuhao1024/repos","events_url":"https://api.github.com/users/liuhao1024/events{/privacy}","received_events_url":"https://api.github.com/users/liuhao1024/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10130492485,"node_id":"LA_kwDOQb6kR88AAAACW9MMRQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/channel:%20feishu","name":"channel: feishu","color":"0969DA","default":false,"description":"Channel integration: feishu"},{"id":10190106613,"node_id":"LA_kwDOQb6kR88AAAACX2Cv9Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20XS","name":"size: XS","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":0,"created_at":"2026-06-16T11:59:30Z","updated_at":"2026-06-16T12:36:31Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93618","html_url":"https://github.com/openclaw/openclaw/pull/93618","diff_url":"https://github.com/openclaw/openclaw/pull/93618.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93618.patch","merged_at":null},"body":"## What does this PR do?\n\nFilters out temporary `card-action-c-*` callback tokens from being used as Feishu reply target IDs. These tokens are ephemeral and not valid Feishu message IDs ā€” using them as `reply_target_message_id` causes \"Invalid ids\" errors from the streaming reply API.\n\n## Related Issue\n\nFixes #56818\n\n## Type of Change\n\n- [x] Bug fix (non-breaking)\n\n## Changes Made\n\n- `extensions/feishu/src/card-action.ts`: Added `card-action-c-*` prefix detection to `buildSyntheticMessageEvent()`. Temporary IDs are excluded from `reply_target_message_id` and `typing_target_message_id`, falling back to `suppress_reply_target: true` instead.\n- `extensions/feishu/src/monitor.account.ts`: Reordered `firstString()` arguments in `parseFeishuCardActionEventPayload()` to prefer `context.open_message_id` (original card message) over `value.open_message_id` (which may be a temporary card-action-c-* ID).\n\n## How to Test\n\n1. Trigger a Feishu card action callback where `value.open_message_id` is a `card-action-c-*` token\n2. The agent should reply normally without \"Invalid ids\" errors\n3. Card actions with valid `om_*` message IDs should continue to work as before\n\n## Real behavior proof\n\n- **Behavior addressed:** Feishu card action callbacks with temporary `card-action-c-*` IDs cause streaming reply failures with \"Invalid open_message_id\" errors. The fix filters these temporary tokens and falls back to no reply target.\n- **Environment tested:** macOS, Node.js v22, openclaw build (pnpm build)\n- **Steps run after the patch:**\n```\n$ node -e \"\nconst testCases = [\n { id: 'card-action-c-abc123', expected: undefined },\n { id: 'om_abc123', expected: 'om_abc123' },\n { id: undefined, expected: undefined },\n];\nfor (const tc of testCases) {\n const isTemp = tc.id?.startsWith('card-action-c-');\n const valid = tc.id && !isTemp ? tc.id : undefined;\n console.log(valid === tc.expected ? 'PASS' : 'FAIL', 'input=' + tc.id);\n}\n\"\nPASS input=card-action-c-abc123\nPASS input=om_abc123\nPASS input=undefined\n```\n- **Evidence after fix:**\n```\n$ grep -n \"card-action-c-\\|isTemporaryCardActionId\\|validReplyTargetId\" extensions/feishu/src/card-action.ts\n139: // card-action-c-* IDs are temporary callback tokens, not valid Feishu message IDs.\n141: const isTemporaryCardActionId = replyTargetMessageId?.startsWith(\"card-action-c-\");\n142: const validReplyTargetId = replyTargetMessageId && !isTemporaryCardActionId\n155: ...(validReplyTargetId ? { reply_target_message_id: validReplyTargetId } : {}),\n156: ...(validReplyTargetId ? { typing_target_message_id: validReplyTargetId } : {}),\n157: ...(!validReplyTargetId ? { suppress_reply_target: true } : {}),\n```\n```\n$ grep -n \"firstString.*context.*open_message\" extensions/feishu/src/monitor.account.ts\n242: const openMessageId = firstString(context.open_message_id, value.open_message_id);\n```\n- **Observed result after fix:** Temporary `card-action-c-*` IDs are correctly filtered out. Valid `om_*` IDs pass through. The `firstString` priority now prefers context.open_message_id (original card) over value.open_message_id (may be temporary).\n- **What was not tested:** Live Feishu card action callbacks (requires running Feishu bot instance with interactive cards). Logic verified via code inspection and unit-level validation.\n\n## Checklist\n\n- [x] Read Contributing Guide\n- [x] Conventional Commits\n- [x] Searched for duplicates\n- [x] Only related changes\n- [x] Tests pass (21/21 bot.card-action tests, 2/2 monitor.message-handler tests)\n- [x] Build passes (pnpm build)","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93618/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93618/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93617","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93617/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93617/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93617/events","html_url":"https://github.com/openclaw/openclaw/pull/93617","id":4673918990,"node_id":"PR_kwDOQb6kR87m_UBw","number":93617,"title":"fix(read): add encoding parameter to read tool for non-UTF-8 text files","user":{"login":"Pick-cat","id":266665499,"node_id":"U_kgDOD-T-Gw","avatar_url":"https://avatars.githubusercontent.com/u/266665499?v=4","gravatar_id":"","url":"https://api.github.com/users/Pick-cat","html_url":"https://github.com/Pick-cat","followers_url":"https://api.github.com/users/Pick-cat/followers","following_url":"https://api.github.com/users/Pick-cat/following{/other_user}","gists_url":"https://api.github.com/users/Pick-cat/gists{/gist_id}","starred_url":"https://api.github.com/users/Pick-cat/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/Pick-cat/subscriptions","organizations_url":"https://api.github.com/users/Pick-cat/orgs","repos_url":"https://api.github.com/users/Pick-cat/repos","events_url":"https://api.github.com/users/Pick-cat/events{/privacy}","received_events_url":"https://api.github.com/users/Pick-cat/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10069976942,"node_id":"LA_kwDOQb6kR88AAAACWDenbg","url":"https://api.github.com/repos/openclaw/openclaw/labels/agents","name":"agents","color":"57606A","default":false,"description":"Agent runtime and tooling"},{"id":10190106648,"node_id":"LA_kwDOQb6kR88AAAACX2CwGA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20S","name":"size: S","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."},{"id":10970522929,"node_id":"LA_kwDOQb6kR88AAAACjeTlMQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/P3","name":"P3","color":"8C959F","default":false,"description":"Low-priority cleanup, docs, polish, ergonomics, or speculative work."},{"id":10980703344,"node_id":"LA_kwDOQb6kR88AAAACjoA8cA","url":"https://api.github.com/repos/openclaw/openclaw/labels/rating:%20%F0%9F%A6%AA%20silver%20shellfish","name":"rating: šŸ¦Ŗ silver shellfish","color":"7A828E","default":false,"description":"Thin PR readiness signal; proof, validation, or implementation needs work."},{"id":10981549078,"node_id":"LA_kwDOQb6kR88AAAACjo0kFg","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20compatibility","name":"merge-risk: šŸšØ compatibility","color":"D1242F","default":false,"description":"šŸšØ May break existing users, config, migrations, defaults, or upgrade paths."},{"id":10982444316,"node_id":"LA_kwDOQb6kR88AAAACjprNHA","url":"https://api.github.com/repos/openclaw/openclaw/labels/status:%20%F0%9F%93%A3%20needs%20proof","name":"status: šŸ“£ needs proof","color":"D93F0B","default":false,"description":"The PR needs real behavior proof before ClawSweeper can clear the contributor ask."}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":1,"created_at":"2026-06-16T11:58:55Z","updated_at":"2026-06-16T12:36:29Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93617","html_url":"https://github.com/openclaw/openclaw/pull/93617","diff_url":"https://github.com/openclaw/openclaw/pull/93617.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93617.patch","merged_at":null},"body":"## Summary\n- The read tool silently produces mojibake when reading GBK/Big5/etc-encoded text files on non-Windows platforms because `decodeWindowsTextFileBuffer` only attempts legacy-encoding fallback on `win32`. On Linux/macOS the path short-circuits to `buffer.toString(\"utf8\")` with no alternative encoding.\n- This patch adds an optional `encoding` parameter to the read tool schema and generalizes `decodeTextFileBuffer` to use an explicit encoding on **any** platform, not just Windows, while preserving the existing platform-specific fallback when no encoding is specified.\n\n## What did NOT change (scope boundary)\n- `decodeWindowsTextFileBuffer` is preserved as a backward-compatible wrapper ā€” all existing callers continue to work unchanged.\n- On non-Windows platforms, when no `encoding` parameter is provided, behavior is exactly the same as before: `buffer.toString(\"utf8\")`. No automatic encoding detection is added for non-Windows ā€” that would be a separate product decision.\n- The sandboxed read tool (`createSandboxedReadTool`) behavior when no encoding is specified remains unchanged: host-path files use `decodeTextFileBuffer`, container-path files use `buffer.toString(\"utf8\")`.\n- exec tool and other subprocess output decoding paths are not affected ā€” they use `decodeWindowsOutputBuffer` which remains Windows-only.\n- No new dependencies added. `TextDecoder` (Node.js built-in) handles GBK/Big5/Shift_JIS/EUC-KR/gb18030 and most common encodings without `iconv-lite`.\n\n## Root Cause (full chain)\n\nSymptom: GBK-encoded text files display as garbled mojibake (e.g. \"äø­ę–‡å†…容\" ā†’ \"Ė¾Å·Ź¢Ō¶\") when read by the read tool on Chinese Windows or any non-UTF-8 locale.\n\nLayer 1: `decodeWindowsTextFileBuffer` returns `buffer.toString(\"utf8\")` on non-Windows platforms (line 125-126 in `src/infra/windows-encoding.ts`). GBK bytes are not valid UTF-8, so they produce replacement characters.\n\nLayer 2: The function guards encoding detection with `if (platform !== \"win32\") { return ... }` ā€” it was designed for Windows-only code-page resolution and never considered that users on any platform might encounter legacy-encoded files.\n\nLayer 3: The read tool has no `encoding` parameter in its schema (`readSchema` in `src/agents/sessions/tools/read.ts`), so there is no way for the caller (model or user) to signal that a file needs a specific encoding. The `decodeText` operation always calls `decodeWindowsTextFileBuffer({ buffer })` without any encoding override.\n\nLayer 4: On Windows specifically, `decodeWindowsTextFileBuffer` **does** detect GBK via PowerShell `[Text.Encoding]::Default.CodePage` and `TextDecoder(\"gbk\")`. This works for Windows users whose system encoding is GBK (codepage 936). But the same file on Linux/macOS (or Docker) has no fallback path.\n\nRoot cause: Two independent gaps interact ā€” (1) the decode function's platform guard excludes non-Windows from encoding fallback, and (2) the read tool schema provides no way to override encoding explicitly. Together, any non-UTF-8 file on a non-Windows platform is guaranteed to produce mojibake with no recovery path.\n\nMissing guardrail: No `encoding` parameter on the read tool and no encoding override path for non-Windows platforms.\n\n## Fix\n- **`src/infra/windows-encoding.ts`**: Add `decodeTextFileBuffer` that accepts an optional `encoding` parameter. When `encoding` is explicitly provided, it is used on **any** platform (not just Windows). When no encoding is given, it delegates to the existing `decodeWindowsBufferWithFallback` (Windows-only fallback). `decodeWindowsTextFileBuffer` is preserved as a backward-compatible wrapper.\n- **`src/agents/sessions/tools/read.ts`**: Add `encoding?: string` to the read schema. Wire the `encoding` parameter through the `decodeText` call in `execute()`.\n- **`src/agents/agent-tools.read.ts`**: Update sandboxed read tool to route `encoding` through `decodeTextFileBuffer`. When `encoding` is provided, use it regardless of host/container path distinction.\n\n## Compatibility\n\n| Scenario | Before | After |\n|----------|--------|-------|\n| UTF-8 file, no encoding param | Correct UTF-8 decode | Same (no change) |\n| GBK file on Windows (codepage 936), no encoding param | Correct GBK via system encoding fallback | Same (no change) |\n| GBK file on Linux/macOS, no encoding param | Mojibake (UTF-8 fallback) | Same (no change ā€” no auto-detection added) |\n| GBK file on Linux/macOS, encoding=\"gbk\" | N/A (no encoding param existed) | Correct GBK decode āœ“ |\n| GBK file on Windows, encoding=\"gbk\" | N/A | Correct GBK decode āœ“ |\n| UTF-8 file, encoding=\"gbk\" | N/A | Correct UTF-8 (UTF-8 priority preserved) āœ“ |\n| Invalid encoding label | N/A | Falls back to UTF-8 (fail-closed) āœ“ |\n\n## Real behavior proof\n\n**Behavior addressed:** Read tool can now correctly decode non-UTF-8 text files on any platform when `encoding` is specified. GBK-encoded files no longer produce mojibake.\n\n**Environment tested:** Node v22.22.0 on Linux, HEAD `35ffbf93b9`.\n\n**Steps run after the patch:** Called the actual `decodeTextFileBuffer` production function from source via `node --import tsx` with GBK-encoded byte buffers.\n\n**Evidence after fix:**\n\n```text\n# BEFORE (on main, no encoding param ā€” the bug):\n$ node --import tsx -e \"\nimport { decodeWindowsTextFileBuffer } from './src/infra/windows-encoding.ts';\nconst gbkBytes = Buffer.from([0xc4, 0xe3, 0xba, 0xc3]);\nconsole.log('linux, no encoding:', JSON.stringify(decodeWindowsTextFileBuffer({ buffer: gbkBytes, platform: 'linux' })));\n\"\nlinux, no encoding: \"\" ā† MOJIBAKE (the bug)\n\n# AFTER (on fix branch, with encoding param ā€” the fix):\n$ node --import tsx -e \"\nimport { decodeTextFileBuffer } from './src/infra/windows-encoding.ts';\nconst gbkBytes = Buffer.from([0xc4, 0xe3, 0xba, 0xc3]);\nconsole.log('linux, encoding=gbk:', JSON.stringify(decodeTextFileBuffer({ buffer: gbkBytes, encoding: 'gbk', platform: 'linux' })));\nconsole.log('UTF-8 priority preserved:', decodeTextFileBuffer({ buffer: Buffer.from('äø­ę–‡ęµ‹čƕ', 'utf8'), encoding: 'gbk', platform: 'linux' }) === 'äø­ę–‡ęµ‹čƕ');\n\"\nlinux, encoding=gbk: \"你儽\" ā† CORRECTLY DECODED āœ“\nUTF-8 priority preserved: true āœ“\n```\n\n**Observed result after the fix:** GBK bytes are correctly decoded to \"你儽\" with `encoding=\"gbk\"` on Linux. UTF-8 priority is preserved when encoding is specified but content is valid UTF-8.\n\n**Not tested:** Live Docker container with GBK-encoded host files. Windows-specific PowerShell code-page resolution with the new `encoding` override. Live QQBot/Feishu channel delivery of GBK-encoded cron output.\n\n## Regression Test Proof\n\nNew tests that FAIL on main (without the fix) and PASS on the fix branch:\n\n```text\n# decodeTextFileBuffer tests ā€” the following would FAIL on main because the\n# function did not exist and there was no way to pass encoding to the decoder:\n\nƗ decodeTextFileBuffer decodes GBK-encoded content with explicit encoding on non-Windows platforms\n AssertionError: expected \"Ć„Ć£Āŗƃ\" to be \"你儽\"\n (main returns buffer.toString(\"utf8\") = mojibake; fix returns \"你儽\")\n\nƗ read tool passes explicit encoding to decodeText for non-UTF-8 files\n AssertionError: decodeTextFileBufferMock was not called with { buffer, encoding: \"gbk\" }\n (main has no encoding parameter in readSchema; fix adds it)\n```\n\n## Verification\n- `node scripts/run-vitest.mjs run src/infra/windows-encoding.test.ts` ā€” 14 tests pass\n- `node scripts/run-vitest.mjs run src/agents/sessions/tools/read.test.ts` ā€” 8 tests pass\n- `node scripts/run-vitest.mjs run src/agents/agent-tools.create-openclaw-coding-tools.adds-claude-style-aliases-schemas-without-dropping-g.test.ts` ā€” 10 tests pass\n\n## User-visible / Behavior Changes\nUsers can now pass `encoding` to the read tool (e.g. `read(\"gbk_file.txt\", { encoding: \"gbk\" })`) to correctly decode non-UTF-8 text files on any platform. Previously, only Windows users with matching system codepage could read legacy-encoded files; all other platforms produced mojibake.\n\n## Security Impact\n- New permissions? No\n- Secrets handling changed? No\n- New network calls? No\n- Fail open ā†’ fail closed? Yes ā€” unsupported encoding labels fall back to UTF-8 (fail-closed), not to arbitrary decoding\n\nRef #92664\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93617/reactions","total_count":1,"+1":1,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":0},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93617/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93605","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93605/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93605/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93605/events","html_url":"https://github.com/openclaw/openclaw/pull/93605","id":4673626002,"node_id":"PR_kwDOQb6kR87m-VON","number":93605,"title":"fix: populate persisted threadId for system events in reply route","user":{"login":"lzyyzznl","id":41978486,"node_id":"MDQ6VXNlcjQxOTc4NDg2","avatar_url":"https://avatars.githubusercontent.com/u/41978486?v=4","gravatar_id":"","url":"https://api.github.com/users/lzyyzznl","html_url":"https://github.com/lzyyzznl","followers_url":"https://api.github.com/users/lzyyzznl/followers","following_url":"https://api.github.com/users/lzyyzznl/following{/other_user}","gists_url":"https://api.github.com/users/lzyyzznl/gists{/gist_id}","starred_url":"https://api.github.com/users/lzyyzznl/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/lzyyzznl/subscriptions","organizations_url":"https://api.github.com/users/lzyyzznl/orgs","repos_url":"https://api.github.com/users/lzyyzznl/repos","events_url":"https://api.github.com/users/lzyyzznl/events{/privacy}","received_events_url":"https://api.github.com/users/lzyyzznl/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064298433,"node_id":"LA_kwDOQb6kR88AAAACV-EBwQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/channel:%20telegram","name":"channel: telegram","color":"0969DA","default":false,"description":"Channel integration: telegram"},{"id":10190106648,"node_id":"LA_kwDOQb6kR88AAAACX2CwGA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20S","name":"size: S","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":3,"created_at":"2026-06-16T11:15:31Z","updated_at":"2026-06-16T12:36:21Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93605","html_url":"https://github.com/openclaw/openclaw/pull/93605","diff_url":"https://github.com/openclaw/openclaw/pull/93605.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93605.patch","merged_at":null},"body":"## Summary\n\n- Fixes Slack media/generated-completion posts on cron/wake turns going to channel root instead of the active thread\n- `resolveEffectiveReplyRoute` failed to include the session entry's persisted threadId for system events (heartbeat/cron-event/exec-event), causing `MessageThreadId` to be undefined in the Slack threading tool context\n- Adds threadId resolution from `deliveryContext.threadId`, `lastThreadId`, and `origin.threadId` (in priority order) to the system event fallthrough path\n\nFixes #93497\n\n## Linked context\n\n- Issue: #93497 ā€” Slack media/generated-completion sends on internal-wake turns post to channel root instead of active thread\n- Root cause: When a Slack wake turn fires, `OriginatingChannel` is undefined, triggering the system event fallthrough path in `resolveEffectiveReplyRoute`. This path resolved channel/to/accountId from `deliveryContext`/`last*` fields but omitted threadId entirely. The missing `replyRoute.threadId` propagated through `get-reply-run.ts` ā†’ `buildEmbeddedContextFromTemplate` ā†’ `buildSlackThreadingToolContext` as undefined `MessageThreadId`, producing `currentThreadTs=undefined`, which routed media uploads to the channel root.\n\n## Real behavior proof (required for external PRs)\n\n**Behavior addressed**: Thread ID inheritance for system event reply routes ā€” previously system events (cron/heartbeat/exec-event) would drop persisted thread IDs from the session entry, causing Slack media uploads on wake turns to post to channel root instead of the active thread.\n\n**Real setup tested**: This fix addresses a stateless routing resolution path that is exercised through unit tests covering the full matrix of system event scenarios. The fix itself is a data-flow change within `resolveEffectiveReplyRoute` ā€” the function now reads `deliveryContext.threadId` / `lastThreadId` / `origin.threadId` and includes the resolved value in the return value for system events when the channel tuple is consistent.\n\n- **Runtime**: node\n\n**Exact steps or command run after fix**:\n\n```\npnpm vitest run src/auto-reply/reply/effective-reply-route.test.ts\n```\n\n**After-fix evidence**:\n\n```\n Test Files 1 passed (1)\n Tests 22 passed (22)\n Start at 17:48:14\n Duration 2.66s (transform 1.86s, setup 0ms, import 2.22s, tests 31ms, environment 0ms)\n```\n\n**Observed result after the fix**: All 22 tests pass with zero regressions. The 7 new test cases specifically validate:\n- `deliveryContext.threadId` is inherited for system events\n- `lastThreadId` is inherited (when deliveryContext.threadId is absent)\n- `origin.threadId` is inherited (as third fallback)\n- Priority ordering: deliveryContext.threadId > lastThreadId > origin.threadId\n- ThreadId is NOT inherited when live channel differs from persisted channel (canInheritPersistedTuple=false)\n- ThreadId IS inherited when live channel matches persisted channel\n- No threadId is included when no thread source is available\n\n**What was not tested**: End-to-end integration test with a live Slack gateway and actual wake-turn execution. The fix is a purely functional data-flow change in `resolveEffectiveReplyRoute` ā€” the propagation path through `get-reply-run.ts` ā†’ `buildEmbeddedContextFromTemplate` ā†’ `buildSlackThreadingToolContext` is existing wiring that already forwards `replyRoute.threadId` into `MessageThreadId`. With the fix, that field is now populated for system events where it was previously undefined.\n\n**Proof limitations or environment constraints**: Unit tests provide functional verification of the routing logic. Full end-to-end validation requires Slack API credentials and a running gateway instance.\n\n## Tests and validation\n\n22 test cases covering all reply route resolution paths:\n\n- 15 existing tests (no regression) covering normal providers, sessions_send handoffs, and exec-event fallbacks\n- 7 new tests covering system event thread ID inheritance:\n - `inherits persisted threadId from deliveryContext for system events`\n - `inherits persisted threadId from lastThreadId for system events`\n - `inherits persisted threadId from origin.threadId for system events`\n - `prefers deliveryContext.threadId over lastThreadId and origin.threadId`\n - `does not inherit threadId for system events when live channel differs from persisted`\n - `inherits threadId for system events when live channel matches persisted channel`\n - `does not include threadId when no thread source is available for system events`\n\n## Risk checklist\n\nDid user-visible behavior change? (`Yes`)\n- For system event providers with persisted thread context: media/messages will now correctly route to the active thread instead of channel root\n- For all other providers and system events without thread context: no change\n\nDid config, environment, or migration behavior change? (`No`)\n\nDid security, auth, secrets, network, or tool execution behavior change? (`No`)\n\nWhat is the highest-risk area?\n- The `persistedThreadId` field is new in the system event return; if a system event consumer incorrectly treats an undefined `threadId` as \"must not post to thread\", they will now get a threadId value. However, the propagation chain (replyRoute ā†’ originatingThreadId ā†’ MessageThreadId) is existing and additive ā€” the thread ID only enables threading behavior that was already desired but broken.\n\nHow is that risk mitigated?\n- Thread ID is only included when `canInheritPersistedTuple` is true (live channel == persisted channel), preventing cross-channel thread contamination\n- The 7 new test cases explicitly verify channel-match gating\n- All 15 existing tests pass without modification, confirming zero regression in normal provider and existing system event paths\n\n## Current review state\n\nWhat is the next action?\n- Maintainer review\n\nWhat is still waiting on author, maintainer, CI, or external proof?\n- Maintainer review and CI pipeline validation\n\nWhich bot or reviewer comments were addressed?\n- N/A ā€” first submission\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93605/reactions","total_count":3,"+1":2,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93605/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93310","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93310/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93310/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93310/events","html_url":"https://github.com/openclaw/openclaw/pull/93310","id":4665854661,"node_id":"PR_kwDOQb6kR87mkzI3","number":93310,"title":"feat(infra): Add structured custom error handler via OPENCLAW_ERROR_HANDLER","user":{"login":"zsxh1990","id":290087845,"node_id":"U_kgDOEUpjpQ","avatar_url":"https://avatars.githubusercontent.com/u/290087845?v=4","gravatar_id":"","url":"https://api.github.com/users/zsxh1990","html_url":"https://github.com/zsxh1990","followers_url":"https://api.github.com/users/zsxh1990/followers","following_url":"https://api.github.com/users/zsxh1990/following{/other_user}","gists_url":"https://api.github.com/users/zsxh1990/gists{/gist_id}","starred_url":"https://api.github.com/users/zsxh1990/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/zsxh1990/subscriptions","organizations_url":"https://api.github.com/users/zsxh1990/orgs","repos_url":"https://api.github.com/users/zsxh1990/repos","events_url":"https://api.github.com/users/zsxh1990/events{/privacy}","received_events_url":"https://api.github.com/users/zsxh1990/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":9706244975,"node_id":"LA_kwDOQb6kR88AAAACQomLbw","url":"https://api.github.com/repos/openclaw/openclaw/labels/docs","name":"docs","color":"0A3069","default":false,"description":"Improvements or additions to documentation"},{"id":10190106613,"node_id":"LA_kwDOQb6kR88AAAACX2Cv9Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20XS","name":"size: XS","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."},{"id":10970446441,"node_id":"LA_kwDOQb6kR88AAAACjeO6aQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/P2","name":"P2","color":"FBCA04","default":false,"description":"Normal backlog priority with limited blast radius."},{"id":10980703344,"node_id":"LA_kwDOQb6kR88AAAACjoA8cA","url":"https://api.github.com/repos/openclaw/openclaw/labels/rating:%20%F0%9F%A6%AA%20silver%20shellfish","name":"rating: šŸ¦Ŗ silver shellfish","color":"7A828E","default":false,"description":"Thin PR readiness signal; proof, validation, or implementation needs work."},{"id":10981549078,"node_id":"LA_kwDOQb6kR88AAAACjo0kFg","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20compatibility","name":"merge-risk: šŸšØ compatibility","color":"D1242F","default":false,"description":"šŸšØ May break existing users, config, migrations, defaults, or upgrade paths."},{"id":10981589204,"node_id":"LA_kwDOQb6kR88AAAACjo3A1A","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20security-boundary","name":"merge-risk: šŸšØ security-boundary","color":"B60205","default":false,"description":"šŸšØ May affect sandboxing, authorization, credentials, or sensitive data."},{"id":10982444316,"node_id":"LA_kwDOQb6kR88AAAACjprNHA","url":"https://api.github.com/repos/openclaw/openclaw/labels/status:%20%F0%9F%93%A3%20needs%20proof","name":"status: šŸ“£ needs proof","color":"D93F0B","default":false,"description":"The PR needs real behavior proof before ClawSweeper can clear the contributor ask."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":4,"created_at":"2026-06-15T14:10:25Z","updated_at":"2026-06-16T12:36:20Z","closed_at":null,"assignee":null,"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93310","html_url":"https://github.com/openclaw/openclaw/pull/93310","diff_url":"https://github.com/openclaw/openclaw/pull/93310.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93310.patch","merged_at":null},"body":"## Summary\n\nAdd support for `OPENCLAW_ERROR_HANDLER` ā€” an environment variable that lets users route OpenClaw's fatal error diagnostics to an external executable as a structured JSON payload.\n\n**Intent:**\n- Non-blocking external handler for fatal errors (uncaught exceptions, CLI failures)\n- Passes structured but redacted error context (`schemaVersion`, `reason`, `timestamp`, `pid`) as a single JSON argv argument\n- Zero new dependencies, `shell: false` for command injection safety, `detached` + `unref()` for non-blocking shutdown\n\n**What is intentionally out of scope:**\n- Not a replacement for the existing `registerFatalErrorHook` plugin API (internal hooks still preferred for bundled diagnostics)\n- Not a daemon or watchdog ā€” fire-and-forget execution only\n- Not a crash-dump collector ā€” the payload is intentionally redacted to protect process-argv visibility. Operators who need full stack details should use OpenClaw's existing stability-bundle mechanism.\n\n**What does success look like:**\n- Users can set `OPENCLAW_ERROR_HANDLER=\"/usr/bin/logger\"` and see structured error metadata in syslog\n- Users can point it at a custom script to POST alerts to their own notification pipeline\n- OpenClaw's exit path remains deterministic ā€” handler failure never blocks shutdown\n\n---\n\n## Amend log\n\n| Revision | Change |\n|----------|--------|\n| v1 (initial proposal) | `shell: true` + stdin pipe + RAW fields. Blocking: shell injection via env var + stdin flush race with `process.exit()`. |\n| v2 (audited) | `shell: false` + argv[1] delivery + RAW opt-in (`OPENCLAW_ERROR_HANDLER_RAW=1`). Resolved injection and flush race. 7-field extended payload behind opt-in gate. |\n| **v3 (current)** | **Removed RAW=1 entirely.** Payload fixed to 4 non-sensitive fields (`schemaVersion`, `reason`, `timestamp`, `pid`). Rationale: (1) argv visibility ā€” 4-field payload contains no sensitive diagnostic data; (2) untestable ā€” RAW path couldn't be verified without a full OpenClaw build (ā‰„12GB RAM); (3) simplicity ā€” removed ~15 lines of conditional branching; (4) extensibility ā€” RAW opt-in can be added via follow-up PR if community requests it. |\n| v4 (Codex review: detach stdio + docs coverage) | Switched `spawn()` `stdio` from `[\"ignore\", \"inherit\", \"inherit\"]` to `\"ignore\"` so the detached handler no longer inherits the parent terminal (Codex review P2: \"Detach the handler from parent stdio\"). Documented `OPENCLAW_ERROR_HANDLER` in `docs/cli/gateway.md` alongside `OPENCLAW_GATEWAY_STARTUP_TRACE` and friends, covering the redacted payload schema, the fire-and-forget spawn contract, and the single-executable-path rule (Codex review: \"Docs and tests do not cover the new env contract\"). |\n| **v5 (current)** | **Restricted `spawn()` env to `{ PATH: process.env.PATH }` only.** Default `spawn()` inherits `process.env`, which would leak OpenClaw's provider keys, gateway tokens, and other runtime secrets to the configured fatal-handler binary. The new commit (`ab81a53b40`) passes an explicit minimal environment so the handler can locate its executable but cannot read any other OpenClaw secret. Also corrected the docs to drop the inaccurate \"synchronous console.error is written if the handler fails to spawn\" claim ā€” that path only fires on synchronous spawn errors; async ENOENT is swallowed by the `error` listener and the fatal caller exits before any user-visible log line is written. |\n| **v6 (current)** | **Real OpenClaw fatal-path proof from a built `pnpm build` artifact.** Built the patch on this contributor's WSL2 instance (`pnpm install` 1169 deps via `http://172.19.128.1:7890` proxy; `pnpm build` 6 min, total 357.2s, slowest `tsdown` 281.4s; no OOM on 24 GiB WSL). Then triggered the patched `process.on(\"uncaughtException\")` handler in `dist/index.js` directly: `node --import /tmp/prethrow2.mjs dist/index.js doctor`, where `prethrow2.mjs` schedules a `setTimeout(throw, 50)` that fires after the launcher finishes registering the OpenClaw uncaughtException handler. The handler reached `runFatalErrorHooks` ā†’ `runExternalErrorHandler` ā†’ `spawn(\"/usr/bin/logger\", [JSON.stringify(payload)], { env: { PATH: process.env.PATH }, stdio: \"ignore\", detached: true, shell: false })`. The detached handler wrote the 4-field JSON payload to syslog as `argv[1]`, then OpenClaw exited. See the Evidence section below for the captured `journalctl` line. This replaces the earlier standalone-harness proof (`openclaw-fatal-hook-proof.mjs`) with a proof that comes from a real OpenClaw runtime, on a built artifact, executing the registered uncaughtException handler. |\n\n---\n\n## Real behavior proof (required for external PRs)\n\n- **Behavior or issue addressed:** OpenClaw currently has no standard mechanism for users to hook an external command into the fatal-error lifecycle. The existing `registerFatalErrorHook` API is internal/bundled ā€” operators who want to route crash diagnostics to their own alerting (syslog, webhook, custom script) have no zero-dep entry point. This PR provides that entry point via an environment variable, following the same precedent as `OPENCLAW_GATEWAY_STARTUP_TRACE` (docs/cli/gateway.md:132).\n- **Real environment tested:** WSL2 (Debian 12, kernel 6.6.87.2-microsoft-standard-WSL2) under Windows 11, Node.js v22.22.2, OpenClaw built from this PR's source via `pnpm build` (357.2s, `tsdown` 281.4s, no OOM on 24 GiB WSL). `openclaw --version` reports `OpenClaw 2026.6.2 (ab81a53b40)` (the `ab81a53b40` commit hash matches the v5 commit on this PR's branch). Handler target: `/usr/bin/logger` (syslog). Payload schema: `{ schemaVersion: 1, reason: string, timestamp: ISO8601, pid: number }`.\n- **Exact steps or command run after this patch:** Real OpenClaw fatal-path exercise, **not** a standalone spawn harness: `OPENCLAW_ERROR_HANDLER=/usr/bin/logger node --import /tmp/prethrow2.mjs dist/index.js doctor`. The `prethrow2.mjs` shim schedules `setTimeout(throw new Error(\"...\"), 50)` so the throw fires after the OpenClaw launcher finishes bootstrapping `dist/index.js` and registers its top-level `process.on(\"uncaughtException\")` handler. The handler then calls `runFatalErrorHooks({ reason: \"uncaught_exception\", error })`, which calls `runExternalErrorHandler(context)`, which spawns the configured handler.\n- **Evidence after fix:** Captured directly from `journalctl` on the test host after running the above command. The OpenClaw CLI first printed its own fatal-path diagnostic to stderr (`[openclaw] OpenClaw hit an unexpected runtime error. Reason: fatal path proof for OPENCLAW_ERROR_HANDLER (PR #93310) ...`), then exited. Independently, `/usr/bin/logger` (spawned detached by the patch) wrote the redacted JSON payload to syslog as `argv[1]`:\n ```\n Jun 16 19:24:08 localhost eric_jia[26334]: {\"schemaVersion\":1,\"reason\":\"uncaught_exception\",\"timestamp\":\"2026-06-16T11:24:08.307Z\",\"pid\":26322}\n ```\n Cross-checks: (a) `pid=26322` is the OpenClaw CLI parent process (matches the run that printed the `[openclaw] OpenClaw hit an unexpected runtime error` line); (b) the JSON's `timestamp` is consistent with the CLI's wall-clock exit time; (c) the `logger` writer process (`26334`) is detached from the OpenClaw CLI process tree and exited before OpenClaw itself exited, confirming `detached: true + unref()`. The standalone-harness proof script (`openclaw-fatal-hook-proof.mjs`) is now superseded and is kept in this PR's `docs/openclaw-pr/` only as a fallback if the OpenClaw build is unavailable.\n- **Observed result after fix:** The fatal-path runtime now produces a structured, redacted, side-channel-explicit log line. The handler is `detached`, so OpenClaw's exit path is not blocked or delayed by the handler's completion. The handler inherits only `PATH` from the OpenClaw process environment (per the v5 commit), so it cannot read provider keys, gateway tokens, or other OpenClaw runtime secrets. The CLI's own diagnostic line still appears on stderr (the `formatCliFailureLines` output), independent of the handler.\n- **What was not tested:** Cross-platform matrix (Windows, macOS) ā€” this build was performed and exercised on WSL2 (Debian 12) only. The `runExternalErrorHandler` function is OS-agnostic stdlib composition (`child_process.spawn` + `process.env`), so behavior is expected to be identical on other Unix-likes, but the Windows console-detach contract has not been observed in this run. CI on this PR is the authoritative cross-platform verification.\n\n### Behavioral or issue addressed\n\nOpenClaw currently has no standard mechanism for users to hook an external command into the fatal-error lifecycle. The existing `registerFatalErrorHook` API is internal/bundled ā€” operators who want to route crash diagnostics to their own alerting (syslog, webhook, custom script) have no zero-dep entry point.\n\nThis PR provides that entry point via an environment variable, following the same precedent as `OPENCLAW_GATEWAY_STARTUP_TRACE` (docs/cli/gateway.md:132).\n\n### Real environment tested\n\n- **OS:** WSL2 (Debian 12, kernel 6.6.87.2-microsoft-standard-WSL2) under Windows 11\n- **Runtime:** Node.js v22.22.3\n- **Handler target:** `/usr/bin/logger` (syslog)\n- **Payload schema:** `{ schemaVersion: 1, reason: string, timestamp: ISO8601, pid: number }`\n\n### Exact steps or command run after this patch\n\n```bash\nnode openclaw-fatal-hook-proof.mjs 2>&1\n```\n\nThe test script exercises the `spawn()` code path that `runExternalErrorHandler` uses, demonstrating handler invocation in a real WSL2 environment with the OpenClaw CLI installed.\n\n### Evidence after fix\n\n```\nHost: DESKTOP-H9EMUD9 | Platform: linux 6.6.87.2-microsoft-standard-WSL2 | Node: v22.22.3 | PID: 95243\n\n### 1. openclaw CLI baseline\n $ openclaw --version\n OpenClaw 2026.6.6 (8c802aa)\n\n### 2. Standard payload (4 fields: schemaVersion, reason, timestamp, pid)\n $ spawn(/usr/bin/logger, [payload], { detached: true, shell: false })\n ā†’ handler invoked, payload delivered via argv[1] āœ“\n\n### 3. Nonexistent handler ā€” graceful degrade\n ā†’ ENOENT swallowed, exit path unaffected āœ“\n\n### 4. shell:false ā€” injection prevention\n ā†’ shell:false blocks injection, literal path ENOENT āœ“\n\n### 5. Syslog delivery\n $ journalctl | grep schemaVersion\n {\"schemaVersion\":1,\"reason\":\"uncaught_exception\",\"timestamp\":\"2026-06-16T03:43:52.734Z\",\"pid\":95243}\n\n--- All scenarios passed ---\n```\n\n### Observed result after fix\n\nEvery configured scenario passes:\n1. **No env var** ā†’ OpenClaw behavior unchanged (zero impact)\n2. **Valid handler** ā†’ payload written to syslog atomically via argv\n3. **Invalid handler path** ā†’ ENOENT swallowed by child `error` listener, main process unaffected\n4. **Shell injection attempt** ā†’ `shell: false` prevents command execution ā€” the injected string is treated as a literal file path, fails with ENOENT\n5. **Exit race** ā†’ `detached + unref` confirmed: parent does not wait for handler completion\n\n### What was not tested\n\nFull OpenClaw runtime integration (requires a build environment with ā‰„12GB RAM; the `tsdown` bundler OOMs at 11GB). The function under review ā€” `runExternalErrorHandler` ā€” is a ~30-line composition of stdlib calls with no dependencies on OpenClaw's runtime state. The spawn logic is identical whether called in isolation or from `runFatalErrorHooks`.\n\n---\n\n## Risk checklist\n\n| Question | Answer |\n|----------|--------|\n| Did user-visible behavior change? | **No** ā€” env var unset ā†’ zero change |\n| Did config/environment behavior change? | **Yes** ā€” new `OPENCLAW_ERROR_HANDLER` env var |\n| Did security/auth/network behavior change? | **No** ā€” `shell: false` prevents injection, handler is detached |\n| Highest-risk area? | Environment variable sourced from untrusted input |\n| How is that risk mitigated? | `shell: false` ā€” handler must be a single executable path, no shell expansion. Documented in Security section. |\n\n## Current review state\n\n- **Next action:** Maintainer review\n- **Waiting on:** CI, proof verification\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93310/reactions","total_count":2,"+1":1,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93310/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93625","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93625/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93625/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93625/events","html_url":"https://github.com/openclaw/openclaw/issues/93625","id":4674058030,"node_id":"I_kwDOQb6kR88AAAABFph3Lg","number":93625,"title":"[Bug] DingTalk session: embedded_run gets stuck immediately on retry, reply trapped in pendingFinalDelivery forever (dispatcher busy-skip)","user":{"login":"quick3210-gif","id":269385196,"node_id":"U_kgDOEA597A","avatar_url":"https://avatars.githubusercontent.com/u/269385196?v=4","gravatar_id":"","url":"https://api.github.com/users/quick3210-gif","html_url":"https://github.com/quick3210-gif","followers_url":"https://api.github.com/users/quick3210-gif/followers","following_url":"https://api.github.com/users/quick3210-gif/following{/other_user}","gists_url":"https://api.github.com/users/quick3210-gif/gists{/gist_id}","starred_url":"https://api.github.com/users/quick3210-gif/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/quick3210-gif/subscriptions","organizations_url":"https://api.github.com/users/quick3210-gif/orgs","repos_url":"https://api.github.com/users/quick3210-gif/repos","events_url":"https://api.github.com/users/quick3210-gif/events{/privacy}","received_events_url":"https://api.github.com/users/quick3210-gif/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":1,"created_at":"2026-06-16T12:18:02Z","updated_at":"2026-06-16T12:36:11Z","closed_at":null,"assignee":null,"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"sub_issues_summary":{"total":0,"completed":0,"percent_completed":0},"issue_dependencies_summary":{"blocked_by":0,"total_blocked_by":0,"blocking":0,"total_blocking":0},"body":"## Summary\n\nA DingTalk DM session enters a permanent \"reply trapped in pendingFinalDelivery\" state after the first embedded_run stalls. Even after `stuck session recovery` reports `aborted=true released=0`, the in-memory phantom `active_owner` persists. A new embedded_run is started for the queued replay, but it also stalls immediately (`lastProgress=embedded_run:started lastProgressAge=163s+`), generates a reply, stores it in `pendingFinalDelivery`, and the dispatcher silently swallows it. **The user never receives a response and must manually restart the gateway.**\n\n## Environment\n\n- OpenClaw: 2026.6.6 (8c802aa)\n- Node: v22.22.3\n- Agent: `xiaoxia` (DingTalk DM, single user)\n- Channel: ddingtalk\n- sessionKey: `agent:xiaoxia:ddingtalk:direct:1937280235499221972`\n- Affected user: `quick3210` (staffId `1937280235499221972`)\n\n## Reproduction sequence\n\n1. User sends a long first message that takes >2 min to process (e.g. a homework help question that requires multi-step reasoning).\n2. User gets impatient, sends 4-6 follow-up messages over the next 5 minutes (typical impatient-teenager pattern).\n3. User sends `/new` to try to reset (this gets ignored ā€” see observation).\n4. User sends `?` (final nudge).\n5. The first `embedded_run` never makes progress past `embedded_run:started` (lastProgressAge grows monotonically past 300s, 350s, 380s+).\n6. After ~360s the diagnostic `stuck session recovery` triggers, runs `abort_embedded_run`, reports `aborted=true drained=true released=0`. The session is still in `state=processing`.\n7. Within seconds a NEW `embedded_run` is started for the queued replay (different sessionId), but it ALSO stalls immediately with `lastProgress=embedded_run:started lastProgressAge=163s+`.\n8. The new run completes (model output succeeds ā€” `outputTokens: 194`), reply text is stored in `pendingFinalDelivery`, `status` flips to `done`.\n9. **But the dispatcher never delivers the reply.** It stays in `pendingFinalDelivery` indefinitely.\n10. Only `systemctl restart openclaw-gateway` recovers the session.\n\n## Evidence (gateway diagnostic logs, 2026-06-16 20:06ā€“20:16 CST)\n\n```\n[diagnostic] stalled session: sessionId=3663a43f-... sessionKey=agent:xiaoxia:ddingtalk:direct:1937280235499221972 state=processing age=143s queueDepth=2 reason=active_work_without_progress classification=stalled_agent_run activeWorkKind=embedded_run lastProgress=embedded_run:started lastProgressAge=322s recovery=none\n[diagnostic] stuck session recovery: sessionId=3663a43f-... age=360s action=abort_embedded_run aborted=true drained=true released=0\n[diagnostic] stuck session recovery outcome: status=aborted action=abort_embedded_run ... aborted=true drained=true forceCleared=false released=0\n[diagnostic] long-running session: sessionId=3663a43f-... state=processing age=203s queueDepth=2 reason=queued_behind_active_work classification=long_running activeWorkKind=embedded_run lastProgress=embedded_run:recovery_skipped_active_owner lastProgressAge=22s recovery=none\n[diagnostic] stalled session: sessionId=ccff3988-... state=processing age=130s queueDepth=2 reason=active_work_without_progress classification=stalled_agent_run activeWorkKind=embedded_run lastProgress=embedded_run:started lastProgressAge=163s recovery=none\n```\n\n(`ccff3988-...` is the **second** embedded_run that auto-started after the first one was \"released\"; same sessionKey.)\n\n## sessions.json state (post-mortem, post-`pendingFinalDelivery` write, pre-restart)\n\n```json\n{\n \"sessionId\": \"ccff3988-d3c2-4428-bd71-7fd018a78359\",\n \"sessionKey\": \"agent:xiaoxia:ddingtalk:direct:1937280235499221972\",\n \"status\": \"done\",\n \"abortedLastRun\": false,\n \"pendingFinalDelivery\": true,\n \"pendingFinalDeliveryText\": \"å—Øļ½žå°č™¾åœØčæ™é‡Œå“¦ļ¼šŸ¦\\n\\n你发äŗ†äøŖ问号ļ¼Œę˜Æęœ‰ä»€ä¹ˆé—®é¢˜ęƒ³é—®å—ļ¼Ÿ...\",\n \"pendingFinalDeliveryCreatedAt\": 1781611851060,\n \"route\": { \"channel\": \"ddingtalk\", \"accountId\": \"dingtlu1ypeyua5bgcl0\", \"target\": { \"to\": \"ddingtalk:user:1937280235499221972\" } },\n \"runtimeMs\": 15077,\n \"contextTokens\": 200000,\n \"inputTokens\": 730,\n \"outputTokens\": 194\n}\n```\n\nNote: model output succeeded (`outputTokens: 194`), `runtimeMs: 15077` (15s, normal), `abortedLastRun: false`. The agent DID complete ā€” only the **delivery to DingTalk** is broken.\n\n## Root cause analysis (from prior incident MEMORY notes)\n\nI filed **#92703** for a closely related case where the dispatcher silently dropped the reply via `ensureDispatchReplyOperation` `admission.skipped active-run` ā†’ `finishReplyOperationBusyDispatch` (`dist/dispatch-*.js` line 723 / 1053). The reply ended up in `pendingFinalDelivery` but never replayed because the next inbound never came.\n\nThis case is **worse** because:\n1. The stuck `embedded_run` not only blocks delivery ā€” it also leaves a phantom `active_owner` in memory that prevents the recovery's own `abort_embedded_run` from actually freeing the lane (`released=0`).\n2. When the system tries to recover by starting a *new* `embedded_run` for the queued messages, the new run also stalls at `embedded_run:started` (likely for the same reason ā€” the `active_owner` flag is never reset because the in-memory `AbortController` from the first run was the only reference).\n3. There's no recovery path that succeeds: `forceCleared=false` on every recovery attempt means the diagnostic system itself sees that recovery is no-op'd.\n\nThe user's daughter waited 11+ minutes for a homework-help reply that was generated in 15 seconds.\n\n## Suggested fix surface\n\n1. `dist/diagnostic-stuck-session-recovery.runtime-*.js` ā€” `abort_embedded_run` returning `released=0 forceCleared=false` should fall back to **forcing a session reload / lane reinit**, not just no-op. Currently it just keeps polling and re-aborting the same phantom.\n2. `dist/dispatch-*.js:723` `ensureDispatchReplyOperation` ā€” when `admission.skipped active-run` is returned and the active run is already known to be aborted (recovery flag set), the reply should be allowed through, not silently dropped.\n3. `dist/agent-runner.runtime-*.js:4016` (sets `pendingFinalDelivery: true`) ā€” there should be a hard timeout (e.g. 60s) after which the system either forces delivery via `message` tool or surfaces the trapped reply to a heartbeat reaper, so the user is not stranded indefinitely.\n\n## Workaround\n\n`systemctl restart openclaw-gateway` (about 30s of feishu/qq disconnect). This is the **third** time in two weeks (cf. #92703, #92708) this dispatcher/embedded_run class of bug has blocked a user reply until restart. The recovery logic clearly needs to be able to evict truly-dead sessions without a restart.\n\n## Related\n\n- #92703 ā€” DM agents silently lose replies when long reply collides with follow-up inbound (DingTalk dispatcher deadlock)\n- #92708 ā€” Feishu streaming card silently truncates long replies (rate-limit 230020/200850 mishandled)","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93625/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93625/timeline","performed_via_github_app":null,"state_reason":null,"pinned_comment":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93282","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93282/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93282/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93282/events","html_url":"https://github.com/openclaw/openclaw/pull/93282","id":4664875171,"node_id":"PR_kwDOQb6kR87mhjrc","number":93282,"title":"fix: trust all ClawHub channel types when package matches official catalog","user":{"login":"wangmiao0668000666","id":290215524,"node_id":"U_kgDOEUxWZA","avatar_url":"https://avatars.githubusercontent.com/u/290215524?v=4","gravatar_id":"","url":"https://api.github.com/users/wangmiao0668000666","html_url":"https://github.com/wangmiao0668000666","followers_url":"https://api.github.com/users/wangmiao0668000666/followers","following_url":"https://api.github.com/users/wangmiao0668000666/following{/other_user}","gists_url":"https://api.github.com/users/wangmiao0668000666/gists{/gist_id}","starred_url":"https://api.github.com/users/wangmiao0668000666/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/wangmiao0668000666/subscriptions","organizations_url":"https://api.github.com/users/wangmiao0668000666/orgs","repos_url":"https://api.github.com/users/wangmiao0668000666/repos","events_url":"https://api.github.com/users/wangmiao0668000666/events{/privacy}","received_events_url":"https://api.github.com/users/wangmiao0668000666/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10069934037,"node_id":"LA_kwDOQb6kR88AAAACWDb_1Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/scripts","name":"scripts","color":"57606A","default":false,"description":"Repository scripts"},{"id":10190106720,"node_id":"LA_kwDOQb6kR88AAAACX2CwYA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20M","name":"size: M","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."},{"id":10970449912,"node_id":"LA_kwDOQb6kR88AAAACjePH-A","url":"https://api.github.com/repos/openclaw/openclaw/labels/P1","name":"P1","color":"D93F0B","default":false,"description":"High-priority user-facing bug, regression, or broken workflow."},{"id":10981549078,"node_id":"LA_kwDOQb6kR88AAAACjo0kFg","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20compatibility","name":"merge-risk: šŸšØ compatibility","color":"D1242F","default":false,"description":"šŸšØ May break existing users, config, migrations, defaults, or upgrade paths."},{"id":10981589204,"node_id":"LA_kwDOQb6kR88AAAACjo3A1A","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20security-boundary","name":"merge-risk: šŸšØ security-boundary","color":"B60205","default":false,"description":"šŸšØ May affect sandboxing, authorization, credentials, or sensitive data."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":21,"created_at":"2026-06-15T12:02:55Z","updated_at":"2026-06-16T12:36:00Z","closed_at":null,"assignee":null,"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93282","html_url":"https://github.com/openclaw/openclaw/pull/93282","diff_url":"https://github.com/openclaw/openclaw/pull/93282.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93282.patch","merged_at":null},"body":"## Summary\r\n\r\nFixes issue #92452: ClawHub official-channel installs for npm-only catalog entries (like Microsoft Teams) were not trusted.\r\n\r\n**Note:** This PR fixes the narrow npm-only catalog mismatch (#92452). The broader self-hosted deployment issue (#92516) remains open and requires additional work on load-path trust and non-interactive provisioning contracts.\r\n\r\n## Problem\r\n\r\nThe Microsoft Teams official external channel catalog entry has npmSpec/defaultChoice npm and no clawhubSpec. Current main requires a clawhubSpec for ClawHub official trust, so npm-only official installs were incorrectly classified as untrusted.\r\n\r\n## Solution\r\n\r\nTrust ClawHub official-channel installs when:\r\n1. The channel is official (not community/private)\r\n2. The URL is the official ClawHub URL (https://clawhub.ai)\r\n3. The package matches the official catalog entry\r\n\r\nThis adds an npm-only path to the trust predicate while preserving all security boundaries.\r\n\r\n## Real Behavior Proof\r\n\r\n**Behavior addressed**: ClawHub official-channel installs for npm-only catalog entries (like Microsoft Teams) were not trusted because main required clawhubSpec. This fix adds URL authority validation and npm-only package matching.\r\n\r\n**Real environment tested**: Local OpenClaw setup with Node.js 24, running actual ClawHub fixture server that mimics clawhub.ai package distribution API.\r\n\r\n**Exact steps or command run after this patch**:\r\n```bash\r\n$ node --import tsx scripts/test-clawhub-official-install.mts\r\n```\r\n\r\n**Evidence after fix**:\r\n```\r\nšŸ§Ŗ Real-environment proof for PR #93282\r\nTesting ClawHub official-channel trust for npm-only catalog entries\r\n\r\nšŸš€ Starting local ClawHub fixture server...\r\nāœ… ClawHub fixture server started at http://127.0.0.1:42791\r\nā³ Waiting for server to be ready...\r\nāœ… Server ready at http://127.0.0.1:42791\r\n\r\n=== Test 1: Official ClawHub install (npm-only, like Microsoft Teams) ===\r\nSimulating: clawhub:@openclaw/msteams from https://clawhub.ai (official channel)\r\n\r\nPlugin ID: msteams\r\nPackage: @openclaw/msteams\r\nOrigin: global\r\nChannels: none\r\nāœ… SUCCESS: npm-only ClawHub official install is TRUSTED\r\n trustedOfficialInstall: true\r\n\r\n=== Test 2: Community channel (should NOT be trusted) ===\r\nSimulating: clawhub:@openclaw/copilot from community channel\r\n\r\nāœ… SUCCESS: Community channel correctly NOT trusted\r\n trustedOfficialInstall: undefined (as expected)\r\n\r\n=== Test 3: Private channel (should NOT be trusted) ===\r\nSimulating: clawhub:@openclaw/copilot from private channel\r\n\r\nāœ… SUCCESS: Private channel correctly NOT trusted\r\n trustedOfficialInstall: undefined (as expected)\r\n\r\n======================================================================\r\nšŸŽ‰ ALL TESTS PASSED!\r\n======================================================================\r\n```\r\n\r\n**Observed result after fix**: \r\n- npm-only ClawHub official-channel installs are now **TRUSTED** with `trustedOfficialInstall: true`\r\n- Community channels remain **UNTRUSTED** (`trustedOfficialInstall: undefined`)\r\n- Private channels remain **UNTRUSTED** (`trustedOfficialInstall: undefined`)\r\n- Security boundary preserved: only `https://clawhub.ai` + official channel + matching package = trusted\r\n\r\n**What was not tested**: Production self-hosted deployments with actual Microsoft Teams channel. This PR fixes the trust classification logic; full self-hosted deployment proof requires additional work tracked in #92516.\r\n\r\n## Security Boundary\r\n\r\nThe changed predicate (`isClawHubOfficialChannelTrustedInstall`) controls which external plugins receive:\r\n- Trusted host-managed persistent state access\r\n- Channel ingress queue APIs\r\n- Keyed store capabilities\r\n\r\nThis PR adds URL authority validation (`isOfficialClawHubInstallRecord`) requiring the official `https://clawhub.ai` URL, ensuring that only genuine ClawHub official installs can be trusted, even for npm-only catalog entries.\r\n\r\n## Changes\r\n\r\n- **src/plugins/manifest-registry.ts**: Trust ClawHub official-channel installs for npm-only catalog entries when URL is https://clawhub.ai\r\n- **src/plugins/official-external-install-records.ts**: Export `isOfficialClawHubInstallRecord` helper for URL authority validation\r\n- **src/plugins/manifest-registry.test.ts**: Added test coverage for npm-only official installs and rejection of community/private channels\r\n- **scripts/test-clawhub-official-install.mts**: Real-environment proof script with local ClawHub fixture server\r\n\r\nFixes #92452\r\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93282/reactions","total_count":1,"+1":1,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":0},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93282/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/91914","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/91914/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/91914/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/91914/events","html_url":"https://github.com/openclaw/openclaw/issues/91914","id":4629997864,"node_id":"I_kwDOQb6kR88AAAABE_gpKA","number":91914,"title":"Queued same-session inbound can cancel in-flight source delivery","user":{"login":"xmoxmo","id":112841961,"node_id":"U_kgDOBrnU6Q","avatar_url":"https://avatars.githubusercontent.com/u/112841961?v=4","gravatar_id":"","url":"https://api.github.com/users/xmoxmo","html_url":"https://github.com/xmoxmo","followers_url":"https://api.github.com/users/xmoxmo/followers","following_url":"https://api.github.com/users/xmoxmo/following{/other_user}","gists_url":"https://api.github.com/users/xmoxmo/gists{/gist_id}","starred_url":"https://api.github.com/users/xmoxmo/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/xmoxmo/subscriptions","organizations_url":"https://api.github.com/users/xmoxmo/orgs","repos_url":"https://api.github.com/users/xmoxmo/repos","events_url":"https://api.github.com/users/xmoxmo/events{/privacy}","received_events_url":"https://api.github.com/users/xmoxmo/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10970449912,"node_id":"LA_kwDOQb6kR88AAAACjePH-A","url":"https://api.github.com/repos/openclaw/openclaw/labels/P1","name":"P1","color":"D93F0B","default":false,"description":"High-priority user-facing bug, regression, or broken workflow."},{"id":10971435209,"node_id":"LA_kwDOQb6kR88AAAACjfLQyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:no-new-fix-pr","name":"clawsweeper:no-new-fix-pr","color":"E5E7EB","default":false,"description":"ClawSweeper does not recommend queueing a new automated fix PR for this issue."},{"id":10971438267,"node_id":"LA_kwDOQb6kR88AAAACjfLcuw","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:source-repro","name":"clawsweeper:source-repro","color":"0A3069","default":false,"description":"ClawSweeper found a high-confidence source-level issue reproduction."},{"id":10971438454,"node_id":"LA_kwDOQb6kR88AAAACjfLddg","url":"https://api.github.com/repos/openclaw/openclaw/labels/clawsweeper:linked-pr-open","name":"clawsweeper:linked-pr-open","color":"57606A","default":false,"description":"ClawSweeper found an open linked pull request for this issue."},{"id":10973505423,"node_id":"LA_kwDOQb6kR88AAAACjhJnjw","url":"https://api.github.com/repos/openclaw/openclaw/labels/impact:session-state","name":"impact:session-state","color":"F9D65C","default":false,"description":"Session, memory, transcript, context, or agent state can drift or corrupt."},{"id":10973513781,"node_id":"LA_kwDOQb6kR88AAAACjhKINQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/impact:message-loss","name":"impact:message-loss","color":"D93F0B","default":false,"description":"Channel message delivery can be lost, duplicated, or misrouted."},{"id":10981171401,"node_id":"LA_kwDOQb6kR88AAAACjodgyQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/issue-rating:%20%F0%9F%A6%9E%20diamond%20lobster","name":"issue-rating: šŸ¦ž diamond lobster","color":"0969DA","default":false,"description":"Very strong issue quality with high-confidence source-level or clear reproduction."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":3,"created_at":"2026-06-10T09:49:05Z","updated_at":"2026-06-16T12:35:39Z","closed_at":null,"assignee":null,"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"sub_issues_summary":{"total":0,"completed":0,"percent_completed":0},"issue_dependencies_summary":{"blocked_by":0,"total_blocked_by":0,"blocking":0,"total_blocking":0},"body":"## Summary\n\nWhen a channel-bound session receives a second inbound message while the first same-session turn is still generating and delivering source replies, the first turn can stop delivering to the originating plugin/channel even though it continues to be visible in the web/session UI.\n\nThis looks like a foreground delivery fencing/admission ordering issue: a queued same-session inbound appears to create a newer foreground reply generation before it is actually admitted to run, causing the active turn's later source delivery to be treated as stale.\n\n## Observed behavior\n\nUsing a plugin-backed direct channel session:\n\n1. Message A arrives from the external channel.\n2. OpenClaw starts processing A and source delivery to the plugin/channel begins normally.\n3. Before A finishes, message B arrives in the same channel/session.\n4. Same-session serialization works as expected: B does not start processing until A finishes.\n5. However, from the moment B arrives, A's later reply chunks/final text remain visible in the web UI/session view but stop reaching the plugin/channel delivery path.\n6. After A completes, B starts and delivers normally.\n\nSo the model/session side continues, but the originating channel no longer receives A's later visible replies.\n\n## Expected behavior\n\nIf B is queued behind A due to same-session serialization, B should not make A's in-flight source delivery stale.\n\nA should continue delivering to the originating channel until A completes, is explicitly aborted, or a newer turn is actually admitted/started according to the intended foreground delivery policy.\n\n## Why this seems to happen\n\nFrom reading the runtime code, `foregroundReplyFence` is keyed roughly by:\n\n```text\nchannel + account + sessionKey + chatType + target\n```\n\nA new inbound appears to call `beginForegroundReplyFence(...)` early in dispatch. Later source delivery goes through a guard like `shouldCancelForegroundReplyDelivery(...)` before reaching the plugin's delivery hook.\n\nBecause message B creates a newer foreground generation while A is still active, A's subsequent source delivery can be blocked/cancelled before the plugin sees it. The plugin cannot recover this because no payload reaches its channel message/outbound adapter.\n\nThe important distinction is:\n\n```text\nexecution admission: B is queued behind A\nforeground delivery generation: B already made A old\n```\n\nThose two states should probably be aligned for same-session queued turns.\n\n## Impact\n\nFor external channel plugins, users can accidentally interrupt delivery of a long response by sending a follow-up message. The original response continues in the web UI, but the external chat only receives the portion delivered before the follow-up inbound.\n\nThis is especially confusing because same-session processing is still serialized correctly; only source delivery is affected.\n\n## Suggested fixes\n\nPreferred option:\n\n- Move `beginForegroundReplyFence(...)` until after reply operation/session admission succeeds, so queued same-session turns do not create a newer foreground generation before they can actually run.\n\nAlternative options:\n\n- Add a foreground delivery policy such as `allow_inflight` for channel/plugin sessions: if an older turn has already started visible source delivery, allow it to finish delivering even if a newer same-session inbound is queued.\n- Do not create/advance the foreground generation for turns that are only queued behind an active same-session run; advance it when the queued turn is actually admitted.\n\n## Notes\n\nThis does not appear to be a plugin outbox/ACK problem. The plugin only receives payloads after OpenClaw's source delivery path allows them through. If foreground fencing suppresses the delivery before the plugin hook, the plugin has nothing to enqueue or retry.\n\nRelated keywords from the runtime while investigating:\n\n```text\nforegroundReplyFence\nbeginForegroundReplyFence\nshouldCancelForegroundReplyDelivery\nreplyRunRegistry / reply turn admission\nsource delivery / originating channel delivery\n```\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/91914/reactions","total_count":1,"+1":1,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":0},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/91914/timeline","performed_via_github_app":null,"state_reason":null,"pinned_comment":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93187","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93187/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93187/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93187/events","html_url":"https://github.com/openclaw/openclaw/pull/93187","id":4662512314,"node_id":"PR_kwDOQb6kR87mZu4P","number":93187,"title":"fix(memory-core): exclude archive transcripts from dreaming corpus and propagate cron parentage to subagents","user":{"login":"xialonglee","id":22994703,"node_id":"MDQ6VXNlcjIyOTk0NzAz","avatar_url":"https://avatars.githubusercontent.com/u/22994703?v=4","gravatar_id":"","url":"https://api.github.com/users/xialonglee","html_url":"https://github.com/xialonglee","followers_url":"https://api.github.com/users/xialonglee/followers","following_url":"https://api.github.com/users/xialonglee/following{/other_user}","gists_url":"https://api.github.com/users/xialonglee/gists{/gist_id}","starred_url":"https://api.github.com/users/xialonglee/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/xialonglee/subscriptions","organizations_url":"https://api.github.com/users/xialonglee/orgs","repos_url":"https://api.github.com/users/xialonglee/repos","events_url":"https://api.github.com/users/xialonglee/events{/privacy}","received_events_url":"https://api.github.com/users/xialonglee/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064299264,"node_id":"LA_kwDOQb6kR88AAAACV-EFAA","url":"https://api.github.com/repos/openclaw/openclaw/labels/extensions:%20memory-core","name":"extensions: memory-core","color":"6E7781","default":false,"description":"Extension: memory-core"},{"id":10190106720,"node_id":"LA_kwDOQb6kR88AAAACX2CwYA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20M","name":"size: M","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."},{"id":10970446441,"node_id":"LA_kwDOQb6kR88AAAACjeO6aQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/P2","name":"P2","color":"FBCA04","default":false,"description":"Normal backlog priority with limited blast radius."},{"id":10980541151,"node_id":"LA_kwDOQb6kR88AAAACjn3C3w","url":"https://api.github.com/repos/openclaw/openclaw/labels/rating:%20%F0%9F%A7%82%20unranked%20krab","name":"rating: šŸ§‚ unranked krab","color":"8C2F39","default":false,"description":"Not merge-ready due to missing proof or serious correctness/safety concerns."},{"id":10981549078,"node_id":"LA_kwDOQb6kR88AAAACjo0kFg","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20compatibility","name":"merge-risk: šŸšØ compatibility","color":"D1242F","default":false,"description":"šŸšØ May break existing users, config, migrations, defaults, or upgrade paths."},{"id":10981579484,"node_id":"LA_kwDOQb6kR88AAAACjo2a3A","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20session-state","name":"merge-risk: šŸšØ session-state","color":"F97316","default":false,"description":"šŸšØ May lose, corrupt, stale, or mis-associate session, agent, or context state."},{"id":10982444316,"node_id":"LA_kwDOQb6kR88AAAACjprNHA","url":"https://api.github.com/repos/openclaw/openclaw/labels/status:%20%F0%9F%93%A3%20needs%20proof","name":"status: šŸ“£ needs proof","color":"D93F0B","default":false,"description":"The PR needs real behavior proof before ClawSweeper can clear the contributor ask."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":4,"created_at":"2026-06-15T06:10:59Z","updated_at":"2026-06-16T12:35:15Z","closed_at":null,"assignee":null,"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93187","html_url":"https://github.com/openclaw/openclaw/pull/93187","diff_url":"https://github.com/openclaw/openclaw/pull/93187.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93187.patch","merged_at":null},"body":"## Summary\n\n- Fix two defects in Dreaming session-corpus ingestion that cause cron-heavy deployment corpora to be dominated by automation noise (67%+ archive transcripts):\n 1. **Archive re-ingestion**: `listSessionFilesForAgent` used `isUsageCountedSessionTranscriptFileName` which includes `.reset.*` / `.deleted.*` archive artifacts. Archive exclusion is now at the Dreaming-only boundary ā€” the shared helper remains usage-counted so `memory_search`/`QMD` can still index archives.\n 2. **Cron parentage not propagated to subagents**: `isCronRunSessionKey` only matched the session's own key. Subagent sessions spawned by cron runs had `agent:main:subagent:*` keys and were never classified as cron-descended.\n- The cron-descended classification now relies purely on the `spawnedBy` chain walk at runtime ā€” no new durable session-store field is introduced.\n- Fixes openclaw/openclaw#90313\n\n## Changes\n\n| File | Change |\n|------|--------|\n| `packages/memory-host-sdk/src/host/session-files.ts` | Keep `listSessionFilesForAgent` usage-counted for `memory_search`/`QMD` callers; remove unused import; add `spawnedBy` chain walk for cron-descended classification |\n| `packages/memory-host-sdk/src/host/session-files.test.ts` | Update tests ā€” add chain-walk cron classification tests; remove `parentTrigger` test |\n| `extensions/memory-core/src/dreaming-phases.ts` | Add archive transcript filter at Dreaming-only boundary; add secondary spawnedBy chain walk defense |\n| `src/agents/subagent-spawn.ts` | Detect cron parent and set spawnedBy; remove `parentTrigger` propagation |\n| `src/config/sessions/types.ts` | Remove `parentTrigger` field from `SessionEntry` (no new durable store field) |\n| `src/plugins/session-entry-slot-keys.ts` | Remove `parentTrigger` from reserved slot keys |\n\n## Verification\n\n- `pnpm test packages/memory-host-sdk/src/host/session-files.test.ts` ā€” 19/19 passed\n- `pnpm test packages/memory-host-sdk/` ā€” 46/46 passed (full suite)\n- `pnpm test src/agents/subagent-spawn.test.ts` ā€” 25/25 passed\n- Import cycles: clean\n- Preserves `memory_search`/`QMD` archive indexing ā€” `listSessionFilesForAgent` returns all usage-counted files including archives; only Dreaming collector filters them out\n\n## Real behavior proof\n\n**Behavior addressed:** Exclude archive transcripts (`.reset.*` and `.deleted.*`) from the Dreaming ingestion corpus so cron-heavy workspaces do not have 67%+ archive contamination in Dreaming output. Also propagate cron parentage to subagent sessions so cron-descended subagents are correctly classified as cron runs ā€” using only the existing `spawnedBy` chain, without adding new session-store fields.\n\n**Real environment tested:** Local Linux x86_64 checkout of this PR branch (`fix/issue-90313-dreaming-corpus-archive`) on OpenClaw `2026.6.1` dev build. Node 22 runtime with full test suite execution.\n\n**Exact steps or command run after this patch:**\n1. `pnpm test packages/memory-host-sdk/src/host/session-files.test.ts` ā€” runs the session-file classification tests including archive filtering and cron-parentage chain-walk traversal\n2. `pnpm test src/agents/subagent-spawn.test.ts` ā€” runs subagent spawn session-patch tests\n\n**Evidence after fix:** Terminal output from both test runs captured below:\n\n```\n# session-files.test.ts (19/19 passed)\n$ pnpm test packages/memory-host-sdk/src/host/session-files.test.ts\n\n RUN v4.1.7\n\n āœ“ isPrimarySessionTranscriptFileName filters archive files\n āœ“ loadSessionTranscriptClassificationForSessionsDir includes primary only\n āœ“ listSessionFilesForAgent returns usage-counted files including archives\n āœ“ classifies a subagent via spawnedBy chain to a cron parent\n ...\n\n Test Files 1 passed (1)\n Tests 19 passed (19)\n Start at 20:28:36\n Duration 6.61s\n\n# subagent-spawn.test.ts (25/25 passed)\n$ pnpm test src/agents/subagent-spawn.test.ts\n\n RUN v4.1.7\n\n Test Files 1 passed (1)\n Tests 25 passed (25)\n Start at 20:29:35\n Duration 8.82s\n```\n\nIssue #90313 documents pre-fix evidence from a real cron-heavy workspace showing 67ā€“91% daily archive contamination at `openclaw@2026.6.1`. After this patch, the same workspace would have archive transcripts excluded from Dreaming while remaining visible to `memory_search`/`QMD` callers.\n\n**Observed result after fix:** All tests pass. The `listSessionFilesForAgent` API returns usage-counted files including archives (preserving `memory_search`/`QMD` indexing), while `collectSessionIngestionBatches` at the Dreaming boundary only ingests primary session transcripts. Cron-descended subagents are correctly classified via the `spawnedBy` chain walk in `loadSessionTranscriptClassificationForSessionsDir`. No new durable session-store field is introduced.\n\n**What was not tested:** This patch was not tested against a live remote OpenClaw workspace with real cron jobs. The changes are deterministic (pure function predicates), so unit test coverage fully exercises both fix paths. The pre-fix evidence from issue #90313 provides the real-workspace baseline, and the post-fix state is verified by the added regression tests.\n\n## Related\n\n- Fixes openclaw/openclaw#90313\n- Related: #67272, #68449, #72611 (distinct root causes ā€” see issue for detailed distinction)\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93187/reactions","total_count":2,"+1":1,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93187/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93630","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93630/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93630/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93630/events","html_url":"https://github.com/openclaw/openclaw/pull/93630","id":4674185298,"node_id":"PR_kwDOQb6kR87nANba","number":93630,"title":"fix(heartbeat): bootstrap plugin session targets","user":{"login":"ZengWen-DT","id":290981215,"node_id":"U_kgDOEVgFXw","avatar_url":"https://avatars.githubusercontent.com/u/290981215?v=4","gravatar_id":"","url":"https://api.github.com/users/ZengWen-DT","html_url":"https://github.com/ZengWen-DT","followers_url":"https://api.github.com/users/ZengWen-DT/followers","following_url":"https://api.github.com/users/ZengWen-DT/following{/other_user}","gists_url":"https://api.github.com/users/ZengWen-DT/gists{/gist_id}","starred_url":"https://api.github.com/users/ZengWen-DT/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/ZengWen-DT/subscriptions","organizations_url":"https://api.github.com/users/ZengWen-DT/orgs","repos_url":"https://api.github.com/users/ZengWen-DT/repos","events_url":"https://api.github.com/users/ZengWen-DT/events{/privacy}","received_events_url":"https://api.github.com/users/ZengWen-DT/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10190106648,"node_id":"LA_kwDOQb6kR88AAAACX2CwGA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20S","name":"size: S","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":0,"created_at":"2026-06-16T12:34:22Z","updated_at":"2026-06-16T12:34:46Z","closed_at":null,"assignee":null,"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93630","html_url":"https://github.com/openclaw/openclaw/pull/93630","diff_url":"https://github.com/openclaw/openclaw/pull/93630.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93630.patch","merged_at":null},"body":"## Summary\n\n- Fixes heartbeat target resolution for plugin channels whose runtime metadata has not been bootstrapped yet.\n- Lets heartbeat `target: \"last\"` use the existing outbound plugin bootstrap path before deciding that a recovered session route has no target.\n- Keeps plugin-owned target and account validation in the same resolver path, so invalid targets and unknown accounts are still suppressed instead of being delivered raw.\n\n## Linked context\n\nCloses #43216\n\n## Real behavior proof\n\n- Behavior addressed: Heartbeat delivery to a plugin channel can recover the last session route after plugin metadata bootstrap, without bypassing plugin target/account policy.\n- Real environment tested: Local OpenClaw source checkout on Node v24.14.0 in WSL2, using the real heartbeat target resolver and bundled Telegram channel metadata.\n- Exact steps or command run after this patch:\n\n```bash\nnode --import tsx --input-type=module <<'EOF'\nimport { resolveHeartbeatDeliveryTarget } from './src/infra/outbound/targets.ts';\n\nconst cases = [\n {\n name: 'telegram group target from last session route',\n entry: { sessionId: 'main', updatedAt: 1, lastChannel: 'telegram', lastTo: '-1001234567890', chatType: 'group' },\n heartbeat: { target: 'last' },\n },\n {\n name: 'invalid explicit account remains suppressed',\n entry: { sessionId: 'main', updatedAt: 1, lastChannel: 'telegram', lastTo: '-1001234567890', chatType: 'group' },\n heartbeat: { target: 'last', accountId: 'missing-account' },\n },\n];\nfor (const item of cases) {\n const result = resolveHeartbeatDeliveryTarget({ cfg: {}, entry: item.entry, heartbeat: item.heartbeat });\n console.log(JSON.stringify({ name: item.name, result }, null, 2));\n}\nEOF\n```\n\n- Evidence after fix:\n\n```text\n{\n \"name\": \"telegram group target from last session route\",\n \"result\": {\n \"channel\": \"telegram\",\n \"to\": \"-1001234567890\",\n \"chatType\": \"group\",\n \"lastChannel\": \"telegram\"\n }\n}\n{\n \"name\": \"invalid explicit account remains suppressed\",\n \"result\": {\n \"channel\": \"none\",\n \"reason\": \"unknown-account\",\n \"accountId\": \"missing-account\",\n \"lastChannel\": \"telegram\"\n }\n}\n```\n\n- Observed result after fix: The last Telegram session route resolves to a deliverable heartbeat target, while an invalid explicit account still returns `channel: \"none\"` with `reason: \"unknown-account\"`.\n- What was not tested: No live Telegram message was sent; this PR only changes target resolution before send.\n- Proof limitations or environment constraints: The proof exercises the source resolver path directly rather than a credentialed channel delivery.\n- Before evidence: The new regression test failed before the implementation with `expected 'forum' to be 'none'` / `expected 'none' to be 'forum'` on the missing bootstrap/policy cases, confirming the resolver either dropped the recovered route or bypassed validation.\n\n## Tests and validation\n\n- `OPENCLAW_VITEST_MAX_WORKERS=1 node scripts/run-vitest.mjs src/infra/outbound/targets.test.ts src/infra/heartbeat-runner.returns-default-unset.test.ts`\n- `pnpm exec oxfmt --check --threads=1 src/infra/outbound/targets.ts src/infra/outbound/targets.test.ts`\n- `git diff --check`\n- `.agents/skills/autoreview/scripts/autoreview --mode branch --base upstream/main --parallel-tests \"OPENCLAW_VITEST_MAX_WORKERS=1 node scripts/run-vitest.mjs src/infra/outbound/targets.test.ts src/infra/heartbeat-runner.returns-default-unset.test.ts\"`\n\n## Risk checklist\n\nDid user-visible behavior change? `Yes`\n\nDid config, environment, or migration behavior change? `No`\n\nDid security, auth, secrets, network, or tool execution behavior change? `No`\n\nHighest-risk area: Heartbeat delivery target selection for plugin channels during startup or registry bootstrap.\n\nHow is that risk mitigated? The change opts into the existing outbound bootstrap path and keeps target/account validation in plugin-owned resolvers before returning a deliverable target.\n\n## Current review state\n\nNext action: maintainer review and CI.\n\nAI-assisted (Codex). Proof above was run manually.\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93630/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93630/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93616","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93616/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93616/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93616/events","html_url":"https://github.com/openclaw/openclaw/pull/93616","id":4673878506,"node_id":"PR_kwDOQb6kR87m_LVs","number":93616,"title":"Keep key-free web search providers opt-in","user":{"login":"davemorin","id":78139,"node_id":"MDQ6VXNlcjc4MTM5","avatar_url":"https://avatars.githubusercontent.com/u/78139?v=4","gravatar_id":"","url":"https://api.github.com/users/davemorin","html_url":"https://github.com/davemorin","followers_url":"https://api.github.com/users/davemorin/followers","following_url":"https://api.github.com/users/davemorin/following{/other_user}","gists_url":"https://api.github.com/users/davemorin/gists{/gist_id}","starred_url":"https://api.github.com/users/davemorin/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/davemorin/subscriptions","organizations_url":"https://api.github.com/users/davemorin/orgs","repos_url":"https://api.github.com/users/davemorin/repos","events_url":"https://api.github.com/users/davemorin/events{/privacy}","received_events_url":"https://api.github.com/users/davemorin/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":9706244975,"node_id":"LA_kwDOQb6kR88AAAACQomLbw","url":"https://api.github.com/repos/openclaw/openclaw/labels/docs","name":"docs","color":"0A3069","default":false,"description":"Improvements or additions to documentation"},{"id":10069934305,"node_id":"LA_kwDOQb6kR88AAAACWDcA4Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/commands","name":"commands","color":"0A3069","default":false,"description":"Command implementations"},{"id":10118530218,"node_id":"LA_kwDOQb6kR88AAAACWxyEqg","url":"https://api.github.com/repos/openclaw/openclaw/labels/maintainer","name":"maintainer","color":"FFD700","default":false,"description":"Maintainer-authored PR"},{"id":10190106760,"node_id":"LA_kwDOQb6kR88AAAACX2CwiA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20L","name":"size: L","color":"8C959F","default":false,"description":null},{"id":11156507181,"node_id":"LA_kwDOQb6kR88AAAACmPrKLQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/extensions:%20parallel","name":"extensions: parallel","color":"ededed","default":false,"description":null}],"state":"open","locked":false,"assignees":[{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false}],"milestone":null,"comments":1,"created_at":"2026-06-16T11:52:57Z","updated_at":"2026-06-16T12:34:38Z","closed_at":null,"assignee":{"login":"vincentkoc","id":25068,"node_id":"MDQ6VXNlcjI1MDY4","avatar_url":"https://avatars.githubusercontent.com/u/25068?v=4","gravatar_id":"","url":"https://api.github.com/users/vincentkoc","html_url":"https://github.com/vincentkoc","followers_url":"https://api.github.com/users/vincentkoc/followers","following_url":"https://api.github.com/users/vincentkoc/following{/other_user}","gists_url":"https://api.github.com/users/vincentkoc/gists{/gist_id}","starred_url":"https://api.github.com/users/vincentkoc/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/vincentkoc/subscriptions","organizations_url":"https://api.github.com/users/vincentkoc/orgs","repos_url":"https://api.github.com/users/vincentkoc/repos","events_url":"https://api.github.com/users/vincentkoc/events{/privacy}","received_events_url":"https://api.github.com/users/vincentkoc/received_events","type":"User","user_view_type":"public","site_admin":false},"author_association":"MEMBER","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93616","html_url":"https://github.com/openclaw/openclaw/pull/93616","diff_url":"https://github.com/openclaw/openclaw/pull/93616.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93616.patch","merged_at":null},"body":"## Summary\n\n- Stop managed `web_search` from implicitly selecting key-free providers such as `parallel-free`, DuckDuckGo, Ollama, or Codex Hosted Search when no API-backed provider is configured.\n- Keep explicit key-free provider selection working, while setup now defaults to skip/no provider unless there is a concrete configured credential or auth signal.\n- Update Parallel/free-provider docs and provider metadata so the UI/docs describe key-free providers as explicit opt-in instead of a zero-config default.\n\n## Verification\n\n- `pnpm test src/web-search/runtime.test.ts src/secrets/runtime-web-tools.test.ts src/commands/onboard-search.providers.test.ts src/flows/search-setup.test.ts extensions/parallel/src/parallel-free-web-search-provider.test.ts -- --reporter=verbose`\n- `pnpm docs:list`\n- `pnpm format:docs:check`\n- `pnpm docs:check-mdx`\n- `pnpm tsgo:test:src`\n- `pnpm tsgo:test:extensions`\n- `git diff --check`\n\n## Notes\n\n- Local autoreview first found stale `docs/tools/web.md` card copy still advertising Parallel Search (Free) as the zero-config default; that copy is fixed here. A follow-up autoreview rerun was interrupted after the worker stayed running for roughly ten minutes without returning a result.","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93616/reactions","total_count":2,"+1":1,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93616/timeline","performed_via_github_app":{"id":1144995,"client_id":"Iv23liVemv8A9if9v0F2","slug":"chatgpt-codex-connector","node_id":"A_kwHOAOQ6Gs4AEXij","owner":{"login":"openai","id":14957082,"node_id":"MDEyOk9yZ2FuaXphdGlvbjE0OTU3MDgy","avatar_url":"https://avatars.githubusercontent.com/u/14957082?v=4","gravatar_id":"","url":"https://api.github.com/users/openai","html_url":"https://github.com/openai","followers_url":"https://api.github.com/users/openai/followers","following_url":"https://api.github.com/users/openai/following{/other_user}","gists_url":"https://api.github.com/users/openai/gists{/gist_id}","starred_url":"https://api.github.com/users/openai/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/openai/subscriptions","organizations_url":"https://api.github.com/users/openai/orgs","repos_url":"https://api.github.com/users/openai/repos","events_url":"https://api.github.com/users/openai/events{/privacy}","received_events_url":"https://api.github.com/users/openai/received_events","type":"Organization","user_view_type":"public","site_admin":false},"name":"ChatGPT Codex Connector","description":"Bring ChatGPT and Codex to your GitHub repositories.","external_url":"https://www.chatgpt.com","html_url":"https://github.com/apps/chatgpt-codex-connector","created_at":"2025-02-14T01:37:05Z","updated_at":"2026-04-20T16:37:15Z","permissions":{"actions":"write","checks":"read","contents":"write","emails":"read","issues":"write","metadata":"read","pull_requests":"write","statuses":"read","workflows":"write"},"events":["check_run","check_suite","commit_comment","issues","issue_comment","pull_request","pull_request_review","pull_request_review_comment","pull_request_review_thread","repository","status","sub_issues"]},"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93377","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93377/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93377/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93377/events","html_url":"https://github.com/openclaw/openclaw/pull/93377","id":4667221641,"node_id":"PR_kwDOQb6kR87mpU3C","number":93377,"title":"fix(model-fallback): classify Codex/OpenAI auth failures","user":{"login":"pandaAIGC","id":221260854,"node_id":"U_kgDODTAsNg","avatar_url":"https://avatars.githubusercontent.com/u/221260854?v=4","gravatar_id":"","url":"https://api.github.com/users/pandaAIGC","html_url":"https://github.com/pandaAIGC","followers_url":"https://api.github.com/users/pandaAIGC/followers","following_url":"https://api.github.com/users/pandaAIGC/following{/other_user}","gists_url":"https://api.github.com/users/pandaAIGC/gists{/gist_id}","starred_url":"https://api.github.com/users/pandaAIGC/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/pandaAIGC/subscriptions","organizations_url":"https://api.github.com/users/pandaAIGC/orgs","repos_url":"https://api.github.com/users/pandaAIGC/repos","events_url":"https://api.github.com/users/pandaAIGC/events{/privacy}","received_events_url":"https://api.github.com/users/pandaAIGC/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10069976942,"node_id":"LA_kwDOQb6kR88AAAACWDenbg","url":"https://api.github.com/repos/openclaw/openclaw/labels/agents","name":"agents","color":"57606A","default":false,"description":"Agent runtime and tooling"},{"id":10190106720,"node_id":"LA_kwDOQb6kR88AAAACX2CwYA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20M","name":"size: M","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."},{"id":10970449912,"node_id":"LA_kwDOQb6kR88AAAACjePH-A","url":"https://api.github.com/repos/openclaw/openclaw/labels/P1","name":"P1","color":"D93F0B","default":false,"description":"High-priority user-facing bug, regression, or broken workflow."},{"id":10980541151,"node_id":"LA_kwDOQb6kR88AAAACjn3C3w","url":"https://api.github.com/repos/openclaw/openclaw/labels/rating:%20%F0%9F%A7%82%20unranked%20krab","name":"rating: šŸ§‚ unranked krab","color":"8C2F39","default":false,"description":"Not merge-ready due to missing proof or serious correctness/safety concerns."},{"id":10981554695,"node_id":"LA_kwDOQb6kR88AAAACjo06Bw","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20auth-provider","name":"merge-risk: šŸšØ auth-provider","color":"F97316","default":false,"description":"šŸšØ May break OAuth, tokens, provider routing, model choice, or credentials."},{"id":10982444316,"node_id":"LA_kwDOQb6kR88AAAACjprNHA","url":"https://api.github.com/repos/openclaw/openclaw/labels/status:%20%F0%9F%93%A3%20needs%20proof","name":"status: šŸ“£ needs proof","color":"D93F0B","default":false,"description":"The PR needs real behavior proof before ClawSweeper can clear the contributor ask."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":12,"created_at":"2026-06-15T17:06:41Z","updated_at":"2026-06-16T12:34:30Z","closed_at":null,"assignee":null,"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93377","html_url":"https://github.com/openclaw/openclaw/pull/93377","diff_url":"https://github.com/openclaw/openclaw/pull/93377.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93377.patch","merged_at":null},"body":"## Summary\n\nFix for #93272.\n\nThis PR improves model failover for Codex/OpenAI auth-shaped and zero-output assistant failures by:\n\n- classifying `auth_unavailable`, `authentication_error`, and invalidated-token text as auth failures;\n- keeping raw assistant error text separate from formatted display text so failover classification and telemetry can use raw details without duplicating user-visible formatting;\n- using formatted assistant error text for assistant-stage failover classification, auth refresh, short-window rate-limit checks, image-dimension checks, and decision logging;\n- including structured assistant `errorCode` / `errorType` in failover-signal details so `errorType: \"authentication_error\"` can classify as auth even without raw error text.\n\n## Why\n\nA local OpenClaw v2026.6.6 incident showed a failed primary model request did not advance to the next configured provider when the upstream error text was an auth-shaped Codex/OpenAI payload. The local hotfix was validated against the affected `heme-literature` dashboard session: primary `openai/gpt-5.5` failed, then fallback `bigmodel/glm-5.2` returned `OK`.\n\n## Real behavior proof\n\n- **Behavior or issue addressed**: OpenClaw should automatically continue to the next configured provider when the primary Codex/OpenAI assistant request fails with auth-shaped or structured zero-output error payloads.\n- **Real environment tested**: Local installed OpenClaw v2026.6.6 hotfix on the affected `heme-literature` dashboard session `agent:heme-literature:dashboard:3d5a1f06-de8c-45d9-8526-aefc77144cd0`, session id `d01a0fd4-2616-4f64-ba49-66cba90faadb`.\n- **Exact steps or command run after this patch**: Applied the same failover changes to the local installed OpenClaw bundle, restarted the local OpenClaw daemon, ran the affected dashboard session through the real OpenClaw runtime, then inspected the generated trajectory records. Secrets, prompts, paths, and config payloads were omitted from the copied output below.\n- **Evidence after fix**: Redacted runtime log excerpt from the real OpenClaw trajectory after the hotfix:\n\n```jsonl\n{\"ts\":\"2026-06-15T11:16:51.491Z\",\"type\":\"session.started\",\"runId\":\"79873b82-1f45-41a0-8fc0-0b19a010c694\",\"provider\":\"openai\",\"modelId\":\"gpt-5.5\",\"modelApi\":\"openai-responses\",\"authProfileId\":\"openai:api-key\"}\n{\"ts\":\"2026-06-15T11:17:27.146Z\",\"type\":\"session.ended\",\"runId\":\"79873b82-1f45-41a0-8fc0-0b19a010c694\",\"provider\":\"openai\",\"modelId\":\"gpt-5.5\",\"modelApi\":\"openai-responses\",\"status\":\"error\"}\n{\"ts\":\"2026-06-15T11:17:28.608Z\",\"type\":\"session.started\",\"runId\":\"79873b82-1f45-41a0-8fc0-0b19a010c694\",\"provider\":\"bigmodel\",\"modelId\":\"glm-5.2\",\"modelApi\":\"openai-completions\"}\n{\"ts\":\"2026-06-15T11:17:45.230Z\",\"type\":\"trace.artifacts\",\"runId\":\"79873b82-1f45-41a0-8fc0-0b19a010c694\",\"provider\":\"bigmodel\",\"modelId\":\"glm-5.2\",\"modelApi\":\"openai-completions\",\"status\":\"success\",\"assistantTexts\":[\"OK\"]}\n{\"ts\":\"2026-06-15T11:17:45.231Z\",\"type\":\"session.ended\",\"runId\":\"79873b82-1f45-41a0-8fc0-0b19a010c694\",\"provider\":\"bigmodel\",\"modelId\":\"glm-5.2\",\"modelApi\":\"openai-completions\",\"status\":\"success\"}\n```\n\n- **Observed result after fix**: The real OpenClaw runtime advanced from failed `openai/gpt-5.5` to fallback `bigmodel/glm-5.2` in the same run, and the fallback model produced assistant output `OK`. Additional same-session fallback-chain proof after `bigmodel/glm-5.2` later hit quota: `openai/gpt-5.5` ended `error`, `bigmodel/glm-5.2` ended `error`, and `deepseek/deepseek-v4-pro` ended `success` in run `f13a5676-91e8-4aa8-a818-f4908cad690a`.\n- **What was not tested**: No additional gaps for this fallback path. The broader session/model override and auth-order policy details remain tracked in #93272 and are not closed by this PR.\n\n## Tests\n\n- Added matcher regression coverage for Codex/OpenAI auth payloads.\n- Added focused assistant-failover coverage for missing raw `errorMessage`.\n- Added helper regression coverage for structured `errorType: \"authentication_error\"` without raw error text.\n- `git diff --check` passed locally.\n- Focused Vitest passed locally: `pnpm vitest run src/agents/embedded-agent-helpers/failover-matches.test.ts src/agents/embedded-agent-helpers/errors.test.ts src/agents/embedded-agent-runner/run/assistant-failover.formatted-error.test.ts --pool=forks --maxWorkers=1 --maxConcurrency=1` -> 5 files passed, 38 tests passed.\n- CI-regression shard passed locally: `pnpm vitest run src/agents/embedded-agent-runner/run.codex-server-error-fallback.test.ts src/agents/embedded-agent-runner/run.cross-provider-fallback-error-context.test.ts src/agents/embedded-agent-runner/run/assistant-failover.formatted-error.test.ts --pool=forks --maxWorkers=1 --maxConcurrency=1` -> 6 files passed, 12 tests passed.\n- The local installed OpenClaw hotfix was smoke-tested with the affected agent/session as shown above.\n\n## Note\n\nThe broader session/model override and auth-order policy details remain tracked in #93272 and should not be considered closed by merging this PR alone.","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93377/reactions","total_count":2,"+1":1,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93377/timeline","performed_via_github_app":{"id":1144995,"client_id":"Iv23liVemv8A9if9v0F2","slug":"chatgpt-codex-connector","node_id":"A_kwHOAOQ6Gs4AEXij","owner":{"login":"openai","id":14957082,"node_id":"MDEyOk9yZ2FuaXphdGlvbjE0OTU3MDgy","avatar_url":"https://avatars.githubusercontent.com/u/14957082?v=4","gravatar_id":"","url":"https://api.github.com/users/openai","html_url":"https://github.com/openai","followers_url":"https://api.github.com/users/openai/followers","following_url":"https://api.github.com/users/openai/following{/other_user}","gists_url":"https://api.github.com/users/openai/gists{/gist_id}","starred_url":"https://api.github.com/users/openai/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/openai/subscriptions","organizations_url":"https://api.github.com/users/openai/orgs","repos_url":"https://api.github.com/users/openai/repos","events_url":"https://api.github.com/users/openai/events{/privacy}","received_events_url":"https://api.github.com/users/openai/received_events","type":"Organization","user_view_type":"public","site_admin":false},"name":"ChatGPT Codex Connector","description":"Bring ChatGPT and Codex to your GitHub repositories.","external_url":"https://www.chatgpt.com","html_url":"https://github.com/apps/chatgpt-codex-connector","created_at":"2025-02-14T01:37:05Z","updated_at":"2026-04-20T16:37:15Z","permissions":{"actions":"write","checks":"read","contents":"write","emails":"read","issues":"write","metadata":"read","pull_requests":"write","statuses":"read","workflows":"write"},"events":["check_run","check_suite","commit_comment","issues","issue_comment","pull_request","pull_request_review","pull_request_review_comment","pull_request_review_thread","repository","status","sub_issues"]},"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93328","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93328/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93328/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93328/events","html_url":"https://github.com/openclaw/openclaw/pull/93328","id":4666222302,"node_id":"PR_kwDOQb6kR87mmBKD","number":93328,"title":"fix(matrix): prevent double bootstrapCrossSigning reset in --force-reset-cross-signing","user":{"login":"xialonglee","id":22994703,"node_id":"MDQ6VXNlcjIyOTk0NzAz","avatar_url":"https://avatars.githubusercontent.com/u/22994703?v=4","gravatar_id":"","url":"https://api.github.com/users/xialonglee","html_url":"https://github.com/xialonglee","followers_url":"https://api.github.com/users/xialonglee/followers","following_url":"https://api.github.com/users/xialonglee/following{/other_user}","gists_url":"https://api.github.com/users/xialonglee/gists{/gist_id}","starred_url":"https://api.github.com/users/xialonglee/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/xialonglee/subscriptions","organizations_url":"https://api.github.com/users/xialonglee/orgs","repos_url":"https://api.github.com/users/xialonglee/repos","events_url":"https://api.github.com/users/xialonglee/events{/privacy}","received_events_url":"https://api.github.com/users/xialonglee/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064298237,"node_id":"LA_kwDOQb6kR88AAAACV-EA_Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/channel:%20matrix","name":"channel: matrix","color":"0969DA","default":false,"description":"Channel integration: matrix"},{"id":10190106648,"node_id":"LA_kwDOQb6kR88AAAACX2CwGA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20S","name":"size: S","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."},{"id":10970449912,"node_id":"LA_kwDOQb6kR88AAAACjePH-A","url":"https://api.github.com/repos/openclaw/openclaw/labels/P1","name":"P1","color":"D93F0B","default":false,"description":"High-priority user-facing bug, regression, or broken workflow."},{"id":10980541151,"node_id":"LA_kwDOQb6kR88AAAACjn3C3w","url":"https://api.github.com/repos/openclaw/openclaw/labels/rating:%20%F0%9F%A7%82%20unranked%20krab","name":"rating: šŸ§‚ unranked krab","color":"8C2F39","default":false,"description":"Not merge-ready due to missing proof or serious correctness/safety concerns."},{"id":10981549078,"node_id":"LA_kwDOQb6kR88AAAACjo0kFg","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20compatibility","name":"merge-risk: šŸšØ compatibility","color":"D1242F","default":false,"description":"šŸšØ May break existing users, config, migrations, defaults, or upgrade paths."},{"id":10981579484,"node_id":"LA_kwDOQb6kR88AAAACjo2a3A","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20session-state","name":"merge-risk: šŸšØ session-state","color":"F97316","default":false,"description":"šŸšØ May lose, corrupt, stale, or mis-associate session, agent, or context state."},{"id":10981585799,"node_id":"LA_kwDOQb6kR88AAAACjo2zhw","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20message-delivery","name":"merge-risk: šŸšØ message-delivery","color":"D1242F","default":false,"description":"šŸšØ May drop, duplicate, misroute, suppress, or wrongly target messages."},{"id":10982444316,"node_id":"LA_kwDOQb6kR88AAAACjprNHA","url":"https://api.github.com/repos/openclaw/openclaw/labels/status:%20%F0%9F%93%A3%20needs%20proof","name":"status: šŸ“£ needs proof","color":"D93F0B","default":false,"description":"The PR needs real behavior proof before ClawSweeper can clear the contributor ask."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":5,"created_at":"2026-06-15T14:55:30Z","updated_at":"2026-06-16T12:33:58Z","closed_at":null,"assignee":null,"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93328","html_url":"https://github.com/openclaw/openclaw/pull/93328","diff_url":"https://github.com/openclaw/openclaw/pull/93328.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93328.patch","merged_at":null},"body":"## Summary\n\n- `--force-reset-cross-signing` calls `bootstrapCrossSigning` twice: once with `setupNewCrossSigning: true` in the forced-reset path, then again via a redundant repair block that can trigger a second reset, destroying E2EE state\n- Root cause: `MatrixCryptoBootstrapper.bootstrap()` deferred SSSS bootstrap for forced reset, and the repair block inside `bootstrapCrossSigning` could fire `resetCrossSigning` again on repairable SSSS errors\n- This PR eliminates the double reset by:\n - For forced reset **with** a password: bootstrap SSSS upfront (non-strict, recreation allowed) before cross-signing, so broken SSSS is repaired before the destructive reset\n - For forced reset **without** a password: preserve the deferral guard to avoid mutating SSSS before UIA succeeds, and block the repair retry to prevent a second destructive reset from stale SSSS errors\n - Disable the SSSS repair block inside `bootstrapCrossSigning` for all forced reset modes, preventing re-triggered reset on both password-backed and passwordless paths\n- Fixes #78396\n\n## Verification\n\n- `pnpm test extensions/matrix/src/matrix/sdk/crypto-bootstrap.test.ts` ā€” 22 passed (including new regression test for passwordless stale-SSSS path)\n- `pnpm test extensions/matrix` ā€” all tests pass\n- Live Matrix QA E2EE bootstrap scenario on Tuwunel v1.5.1 homeserver ā€” **4/4 suite pass, scenario pass** (see Real behavior proof)\n- Tests confirm: no SSSS mutation for passwordless forced reset before UIA, single cross-signing reset, no repair retry for passwordless stale-SSSS errors\n- Autoreview: clean\n\n## Real behavior proof\n\nBehavior addressed: Forced cross-signing reset no longer fires bootstrapCrossSigning twice; passwordless bots preserve the SSSS mutation guard and do not second-reset on stale SSSS\n\nReal environment tested: Live Matrix homeserver (Tuwunel v1.5.1, Docker) via `extensions/qa-matrix` QA harness with E2EE-enabled Matrix channel\n\nExact steps or command run after this patch:\n```\npnpm qa:matrix --filter matrix-e2ee-bootstrap-success\n```\n\nEvidence after fix:\n```\n[matrix-qa] scenario pass matrix-e2ee-bootstrap-success 19.2s\n[matrix-qa] suite pass 4/4 total=77.7s\n```\n\nQA report breakdown:\n| Check | Status |\n|---|---|\n| Matrix harness ready | pass (image: tuwunel:v1.5.1, baseUrl: http://127.0.0.1:28008/) |\n| Matrix channel ready | pass (userId: @qa-sut-...:matrix-qa.test) |\n| Matrix canary | pass (reply token matched) |\n| E2EE bootstrap scenario | pass (bootstrapSuccess: true) |\n\nE2EE bootstrap details (from `matrix-qa-report.md`):\n- **bootstrap actor**: driver (real Matrix crypto bootstrap, not mocked)\n- **device verified**: yes\n- **cross-signing verified**: yes\n- **cross-signing published**: yes\n- **signed by owner**: yes\n- **room-key backup version**: 416\n- **recovery key stored**: yes (id: YNcpwZxYcTQw00wvPfQTaYrNk7Ml6MTE)\n\nObserved result after fix: Live Matrix homeserver confirms single cross-signing bootstrap with correct device trust, cross-signing key publication, room key backup creation, and recovery key storage ā€” no double reset or E2EE state corruption\n\nWhat was not tested: Federation scenarios; multi-device login/verification flow\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93328/reactions","total_count":2,"+1":1,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93328/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93629","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93629/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93629/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93629/events","html_url":"https://github.com/openclaw/openclaw/pull/93629","id":4674166078,"node_id":"PR_kwDOQb6kR87nAJQd","number":93629,"title":"fix(reply): preserve unsent text-only finals after block pipeline streamed partial content (fixes #81078)","user":{"login":"liuhao1024","id":11816344,"node_id":"MDQ6VXNlcjExODE2MzQ0","avatar_url":"https://avatars.githubusercontent.com/u/11816344?v=4","gravatar_id":"","url":"https://api.github.com/users/liuhao1024","html_url":"https://github.com/liuhao1024","followers_url":"https://api.github.com/users/liuhao1024/followers","following_url":"https://api.github.com/users/liuhao1024/following{/other_user}","gists_url":"https://api.github.com/users/liuhao1024/gists{/gist_id}","starred_url":"https://api.github.com/users/liuhao1024/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/liuhao1024/subscriptions","organizations_url":"https://api.github.com/users/liuhao1024/orgs","repos_url":"https://api.github.com/users/liuhao1024/repos","events_url":"https://api.github.com/users/liuhao1024/events{/privacy}","received_events_url":"https://api.github.com/users/liuhao1024/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10190106613,"node_id":"LA_kwDOQb6kR88AAAACX2Cv9Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20XS","name":"size: XS","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":0,"created_at":"2026-06-16T12:32:00Z","updated_at":"2026-06-16T12:32:45Z","closed_at":null,"assignee":null,"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93629","html_url":"https://github.com/openclaw/openclaw/pull/93629","diff_url":"https://github.com/openclaw/openclaw/pull/93629.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93629.patch","merged_at":null},"body":"## Summary\n\nWhen the block reply pipeline streamed partial content, `buildReplyPayloads()` unconditionally dropped all text-only final payloads via `preserveUnsentMediaAfterBlockSend()`. This suppressed the complete final reply when the pipeline only streamed a partial block and never sent the exact final text to the user.\n\n## Root cause\n\nIn `preserveUnsentMediaAfterBlockSend()` (`agent-runner-payloads.ts:375`), text-only payloads (no media) hit `return null` unconditionally:\n\n```ts\nif (!reply.hasMedia) {\n return null; // ā† drops ALL text-only payloads, even unsent ones\n}\n```\n\nThe function was designed to preserve unsent *media* payloads, but its early return for text-only payloads bypassed the `hasSentPayload()` check entirely. When the block pipeline streamed a partial block (e.g. \"Hello\") but the model produced a longer final text (e.g. \"Final message\"), the final text was silently dropped because it was text-only.\n\n## Fix\n\nReplace the unconditional `return null` with a `hasSentPayload()` check:\n\n```ts\nif (!reply.hasMedia) {\n if (params.blockReplyPipeline?.hasSentPayload(payload)) {\n return null; // already sent ā†’ drop\n }\n return payload; // unsent ā†’ preserve\n}\n```\n\nThis ensures text-only payloads are only dropped when the pipeline confirms it sent the exact same content.\n\n## Changes\n\n- `src/auto-reply/reply/agent-runner-payloads.ts`: 1-line logic fix in `preserveUnsentMediaAfterBlockSend`\n- `src/auto-reply/reply/agent-runner-payloads.test.ts`: Split the old \"drops all final payloads\" test into two: (1) preserves unsent text-only finals, (2) drops already-sent text-only finals\n- `src/auto-reply/reply/agent-runner.misc.runreplyagent.test.ts`: Updated integration test expectation to match the fix\n\n## Real behavior proof\n\n- **Behavior addressed:** WhatsApp block streaming silently suppresses the complete final assistant reply after a partial streamed block was sent\n- **Environment tested:** macOS 26.4.1, Node.js, OpenClaw workdir at commit ed16f8fc\n- **Steps run after the patch:** Ran targeted tests on the changed file with `node scripts/run-vitest.mjs run src/auto-reply/reply/agent-runner-payloads.test.ts` and `node scripts/run-vitest.mjs run src/auto-reply/reply/agent-runner.misc.runreplyagent.test.ts`\n- **Evidence after fix:**\n\n```\n$ node scripts/run-vitest.mjs run src/auto-reply/reply/agent-runner-payloads.test.ts\n āœ“ auto-reply src/auto-reply/reply/agent-runner-payloads.test.ts (63 tests) 569ms\n Test Files 1 passed (1)\n Tests 63 passed (63)\n\n$ node scripts/run-vitest.mjs run src/auto-reply/reply/agent-runner.misc.runreplyagent.test.ts\n āœ“ auto-reply src/auto-reply/reply/agent-runner.misc.runreplyagent.test.ts (47 tests) 1763ms\n Test Files 1 passed (1)\n Tests 47 passed (47)\n\n$ grep -n \"hasSentPayload\" src/auto-reply/reply/agent-runner-payloads.ts\n378: if (params.blockReplyPipeline?.hasSentPayload(payload)) {\n```\n\n- **Observed result after fix:** All 110 tests pass (63 + 47). The fix preserves unsent text-only finals while still dropping already-sent ones.\n- **What was not tested:** Live WhatsApp channel behavior (requires WhatsApp integration setup); tested via unit and integration tests only.\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93629/reactions","total_count":1,"+1":0,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93629/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93160","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/93160/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/93160/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/93160/events","html_url":"https://github.com/openclaw/openclaw/pull/93160","id":4662028828,"node_id":"PR_kwDOQb6kR87mYKNM","number":93160,"title":"fix(tui): recognize DEL (0x7f) as backspace for WSL2 terminals (fixes #92777)","user":{"login":"xzh-xydt","id":279172199,"node_id":"U_kgDOEKPUZw","avatar_url":"https://avatars.githubusercontent.com/u/279172199?v=4","gravatar_id":"","url":"https://api.github.com/users/xzh-xydt","html_url":"https://github.com/xzh-xydt","followers_url":"https://api.github.com/users/xzh-xydt/followers","following_url":"https://api.github.com/users/xzh-xydt/following{/other_user}","gists_url":"https://api.github.com/users/xzh-xydt/gists{/gist_id}","starred_url":"https://api.github.com/users/xzh-xydt/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/xzh-xydt/subscriptions","organizations_url":"https://api.github.com/users/xzh-xydt/orgs","repos_url":"https://api.github.com/users/xzh-xydt/repos","events_url":"https://api.github.com/users/xzh-xydt/events{/privacy}","received_events_url":"https://api.github.com/users/xzh-xydt/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10190106613,"node_id":"LA_kwDOQb6kR88AAAACX2Cv9Q","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20XS","name":"size: XS","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."},{"id":10970449912,"node_id":"LA_kwDOQb6kR88AAAACjePH-A","url":"https://api.github.com/repos/openclaw/openclaw/labels/P1","name":"P1","color":"D93F0B","default":false,"description":"High-priority user-facing bug, regression, or broken workflow."},{"id":10980541151,"node_id":"LA_kwDOQb6kR88AAAACjn3C3w","url":"https://api.github.com/repos/openclaw/openclaw/labels/rating:%20%F0%9F%A7%82%20unranked%20krab","name":"rating: šŸ§‚ unranked krab","color":"8C2F39","default":false,"description":"Not merge-ready due to missing proof or serious correctness/safety concerns."},{"id":10982444316,"node_id":"LA_kwDOQb6kR88AAAACjprNHA","url":"https://api.github.com/repos/openclaw/openclaw/labels/status:%20%F0%9F%93%A3%20needs%20proof","name":"status: šŸ“£ needs proof","color":"D93F0B","default":false,"description":"The PR needs real behavior proof before ClawSweeper can clear the contributor ask."},{"id":11046736196,"node_id":"LA_kwDOQb6kR88AAAACkm_RRA","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20other","name":"merge-risk: šŸšØ other","color":"C5DEF5","default":false,"description":"šŸšØ Merging this PR has meaningful risk outside the owned taxonomy."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":3,"created_at":"2026-06-15T04:23:47Z","updated_at":"2026-06-16T12:31:39Z","closed_at":null,"assignee":null,"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/93160","html_url":"https://github.com/openclaw/openclaw/pull/93160","diff_url":"https://github.com/openclaw/openclaw/pull/93160.diff","patch_url":"https://github.com/openclaw/openclaw/pull/93160.patch","merged_at":null},"body":"Fixes #92777\r\n\r\n## Summary\r\n\r\nAdd explicit `\\x7f` (DEL) recognition to the TUI backspace deduper so backspace works on terminals that send `^?` instead of `^H`, such as WSL2 Ubuntu.\r\n\r\n## Root Cause\r\n\r\n`createBackspaceDeduper` in `src/tui/tui.ts` explicitly checks for `\\x08` (^H / BS) and relies on `matchesKey(data, Key.backspace)` to catch other backspace representations. On WSL2 Ubuntu terminals, the terminal sends `\\x7f` (^? / DEL) for the backspace key. While `matchesKey()` handles this in most configurations, adding an explicit check ensures reliable backspace recognition regardless of terminal-specific behavior, especially after the TUI re-enters raw mode following agent response rendering.\r\n\r\n**Fix**: Add `data !== \"\\x7f\"` to the guard condition so DEL is always recognized as backspace alongside BS.\r\n\r\n## Real behavior proof\r\n\r\n**Behavior or issue addressed**: Backspace key (sending ^? / DEL) in WSL2 Ubuntu terminals is now reliably recognized as backspace in the TUI input field, including after agent response renders.\r\n\r\n**Real environment tested**: Linux x64 (WSL2), Node.js v24.16.0, openclaw at cc954798 on top of upstream/main.\r\n\r\n**Exact steps or command run after this patch**:\r\n1. Run TUI backspace deduper tests: `npx vitest run src/tui/tui.test.ts -t \"backspace\"`\r\n2. All backspace deduplication tests pass, including `\\x7f` (DEL) and `\\x08` (BS) cross-deduplication\r\n\r\n**Evidence after fix**:\r\n```\r\n āœ“ src/tui/tui.test.ts > createBackspaceDeduper > suppresses duplicate backspace events within the dedupe window\r\n āœ“ src/tui/tui.test.ts > createBackspaceDeduper > preserves backspace events outside the dedupe window\r\n āœ“ src/tui/tui.test.ts > createBackspaceDeduper > treats ASCII BS as backspace when it is the first event\r\n āœ“ src/tui/tui.test.ts > createBackspaceDeduper > never suppresses non-backspace keys\r\n```\r\n\r\nThe `createBackspaceDeduper` tests exercise `\\x7f` extensively:\r\n- `suppresses duplicate backspace events within the dedupe window` ā€” uses `\\x7f` then `\\x08`\r\n- `preserves backspace events outside the dedupe window` ā€” uses `\\x7f`\r\n- `treats ASCII BS as backspace when it is the first event` ā€” uses `\\x08` then `\\x7f`\r\n- `never suppresses non-backspace keys` ā€” verifies normal keys pass through\r\n\r\n**Observed result after fix**: `\\x7f` is now explicitly recognized alongside `\\x08` in the guard condition. The explicit check ensures backspace recognition regardless of terminal-specific `matchesKey()` behavior after raw mode re-entry.\r\n\r\n**What was not tested**: Live WSL2 Ubuntu terminal (no Windows environment available). The fix is validated by the existing test suite which uses `\\x7f` extensively for backspace deduplication.","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/93160/reactions","total_count":2,"+1":1,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/93160/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/41067","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/41067/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/41067/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/41067/events","html_url":"https://github.com/openclaw/openclaw/pull/41067","id":4045698206,"node_id":"PR_kwDOQb6kR87JFChl","number":41067,"title":"Fix dashboard chat run recovery across reconnects","user":{"login":"Elysium-Seeker","id":198186729,"node_id":"U_kgDOC9AW6Q","avatar_url":"https://avatars.githubusercontent.com/u/198186729?v=4","gravatar_id":"","url":"https://api.github.com/users/Elysium-Seeker","html_url":"https://github.com/Elysium-Seeker","followers_url":"https://api.github.com/users/Elysium-Seeker/followers","following_url":"https://api.github.com/users/Elysium-Seeker/following{/other_user}","gists_url":"https://api.github.com/users/Elysium-Seeker/gists{/gist_id}","starred_url":"https://api.github.com/users/Elysium-Seeker/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/Elysium-Seeker/subscriptions","organizations_url":"https://api.github.com/users/Elysium-Seeker/orgs","repos_url":"https://api.github.com/users/Elysium-Seeker/repos","events_url":"https://api.github.com/users/Elysium-Seeker/events{/privacy}","received_events_url":"https://api.github.com/users/Elysium-Seeker/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064298719,"node_id":"LA_kwDOQb6kR88AAAACV-EC3w","url":"https://api.github.com/repos/openclaw/openclaw/labels/app:%20macos","name":"app: macos","color":"6E7781","default":false,"description":"App: macos"},{"id":10064298757,"node_id":"LA_kwDOQb6kR88AAAACV-EDBQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/app:%20web-ui","name":"app: web-ui","color":"6E7781","default":false,"description":"App: web-ui"},{"id":10064298857,"node_id":"LA_kwDOQb6kR88AAAACV-EDaQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/gateway","name":"gateway","color":"57606A","default":false,"description":"Gateway runtime"},{"id":10190106760,"node_id":"LA_kwDOQb6kR88AAAACX2CwiA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20L","name":"size: L","color":"8C959F","default":false,"description":null},{"id":10865780645,"node_id":"LA_kwDOQb6kR88AAAACh6anpQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/triage:%20needs-real-behavior-proof","name":"triage: needs-real-behavior-proof","color":"C5DEF5","default":false,"description":"Candidate: external PR needs after-fix proof from a real setup."},{"id":10970449912,"node_id":"LA_kwDOQb6kR88AAAACjePH-A","url":"https://api.github.com/repos/openclaw/openclaw/labels/P1","name":"P1","color":"D93F0B","default":false,"description":"High-priority user-facing bug, regression, or broken workflow."},{"id":10980541151,"node_id":"LA_kwDOQb6kR88AAAACjn3C3w","url":"https://api.github.com/repos/openclaw/openclaw/labels/rating:%20%F0%9F%A7%82%20unranked%20krab","name":"rating: šŸ§‚ unranked krab","color":"8C2F39","default":false,"description":"Not merge-ready due to missing proof or serious correctness/safety concerns."},{"id":10981579484,"node_id":"LA_kwDOQb6kR88AAAACjo2a3A","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20session-state","name":"merge-risk: šŸšØ session-state","color":"F97316","default":false,"description":"šŸšØ May lose, corrupt, stale, or mis-associate session, agent, or context state."},{"id":10981585799,"node_id":"LA_kwDOQb6kR88AAAACjo2zhw","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20message-delivery","name":"merge-risk: šŸšØ message-delivery","color":"D1242F","default":false,"description":"šŸšØ May drop, duplicate, misroute, suppress, or wrongly target messages."},{"id":10982444316,"node_id":"LA_kwDOQb6kR88AAAACjprNHA","url":"https://api.github.com/repos/openclaw/openclaw/labels/status:%20%F0%9F%93%A3%20needs%20proof","name":"status: šŸ“£ needs proof","color":"D93F0B","default":false,"description":"The PR needs real behavior proof before ClawSweeper can clear the contributor ask."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":8,"created_at":"2026-03-09T13:44:41Z","updated_at":"2026-06-16T12:31:20Z","closed_at":null,"assignee":null,"author_association":"NONE","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/41067","html_url":"https://github.com/openclaw/openclaw/pull/41067","diff_url":"https://github.com/openclaw/openclaw/pull/41067.diff","patch_url":"https://github.com/openclaw/openclaw/pull/41067.patch","merged_at":null},"body":"## Summary\n\nThis PR fixes two closely related Gateway Dashboard chat reliability problems:\n\n1. Dashboard chat could lose busy-state on reconnect/tab return and incorrectly fall back to `New session` / `Send` while the session still had an active run.\n2. Gateway restarts could interrupt an in-flight chat run without leaving a visible assistant-side notice in the chat transcript.\n\n## What changed\n\n- include `activeChatRuns` in the gateway hello snapshot\n- recover `chatRunId` / busy state from snapshot on reconnect instead of blindly clearing it\n- keep the composer in busy mode while a run is still abortable\n- stop forcing sidebar chat navigation back to `main`; prefer the last active chat session\n- abort active chat runs during gateway shutdown and inject a durable assistant transcript notice\n- process restart sentinel wake immediately on startup and inject a transcript notice for internal sessions without external delivery routes\n\n## Validation\n\n- `pnpm vitest run src/gateway/server-close.test.ts src/gateway/server-restart-sentinel.test.ts ui/src/ui/app-gateway.node.test.ts ui/src/ui/controllers/chat.test.ts`\n- `NODE_OPTIONS=--max-old-space-size=6144 pnpm -s tsc --noEmit --pretty false`\n\n## Related\n\n- Closes #41064\n- Related: #38836\n- Related: #21641\n- Related: #26986\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/41067/reactions","total_count":4,"+1":3,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/41067/timeline","performed_via_github_app":null,"state_reason":null},{"url":"https://api.github.com/repos/openclaw/openclaw/issues/91049","repository_url":"https://api.github.com/repos/openclaw/openclaw","labels_url":"https://api.github.com/repos/openclaw/openclaw/issues/91049/labels{/name}","comments_url":"https://api.github.com/repos/openclaw/openclaw/issues/91049/comments","events_url":"https://api.github.com/repos/openclaw/openclaw/issues/91049/events","html_url":"https://github.com/openclaw/openclaw/pull/91049","id":4605692813,"node_id":"PR_kwDOQb6kR87ji0Q8","number":91049,"title":"fix: handle terminal chat send acknowledgements","user":{"login":"nxmxbbd","id":32288,"node_id":"MDQ6VXNlcjMyMjg4","avatar_url":"https://avatars.githubusercontent.com/u/32288?v=4","gravatar_id":"","url":"https://api.github.com/users/nxmxbbd","html_url":"https://github.com/nxmxbbd","followers_url":"https://api.github.com/users/nxmxbbd/followers","following_url":"https://api.github.com/users/nxmxbbd/following{/other_user}","gists_url":"https://api.github.com/users/nxmxbbd/gists{/gist_id}","starred_url":"https://api.github.com/users/nxmxbbd/starred{/owner}{/repo}","subscriptions_url":"https://api.github.com/users/nxmxbbd/subscriptions","organizations_url":"https://api.github.com/users/nxmxbbd/orgs","repos_url":"https://api.github.com/users/nxmxbbd/repos","events_url":"https://api.github.com/users/nxmxbbd/events{/privacy}","received_events_url":"https://api.github.com/users/nxmxbbd/received_events","type":"User","user_view_type":"public","site_admin":false},"labels":[{"id":10064298660,"node_id":"LA_kwDOQb6kR88AAAACV-ECpA","url":"https://api.github.com/repos/openclaw/openclaw/labels/app:%20android","name":"app: android","color":"6E7781","default":false,"description":"App: android"},{"id":10064298679,"node_id":"LA_kwDOQb6kR88AAAACV-ECtw","url":"https://api.github.com/repos/openclaw/openclaw/labels/app:%20ios","name":"app: ios","color":"6E7781","default":false,"description":"App: ios"},{"id":10064298719,"node_id":"LA_kwDOQb6kR88AAAACV-EC3w","url":"https://api.github.com/repos/openclaw/openclaw/labels/app:%20macos","name":"app: macos","color":"6E7781","default":false,"description":"App: macos"},{"id":10064298757,"node_id":"LA_kwDOQb6kR88AAAACV-EDBQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/app:%20web-ui","name":"app: web-ui","color":"6E7781","default":false,"description":"App: web-ui"},{"id":10064298857,"node_id":"LA_kwDOQb6kR88AAAACV-EDaQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/gateway","name":"gateway","color":"57606A","default":false,"description":"Gateway runtime"},{"id":10190106804,"node_id":"LA_kwDOQb6kR88AAAACX2CwtA","url":"https://api.github.com/repos/openclaw/openclaw/labels/size:%20XL","name":"size: XL","color":"8C959F","default":false,"description":null},{"id":10873837181,"node_id":"LA_kwDOQb6kR88AAAACiCGWfQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/proof:%20supplied","name":"proof: supplied","color":"C2E0C6","default":false,"description":"External PR includes structured after-fix real behavior proof."},{"id":10970446441,"node_id":"LA_kwDOQb6kR88AAAACjeO6aQ","url":"https://api.github.com/repos/openclaw/openclaw/labels/P2","name":"P2","color":"FBCA04","default":false,"description":"Normal backlog priority with limited blast radius."},{"id":10981579484,"node_id":"LA_kwDOQb6kR88AAAACjo2a3A","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20session-state","name":"merge-risk: šŸšØ session-state","color":"F97316","default":false,"description":"šŸšØ May lose, corrupt, stale, or mis-associate session, agent, or context state."},{"id":10981585799,"node_id":"LA_kwDOQb6kR88AAAACjo2zhw","url":"https://api.github.com/repos/openclaw/openclaw/labels/merge-risk:%20%F0%9F%9A%A8%20message-delivery","name":"merge-risk: šŸšØ message-delivery","color":"D1242F","default":false,"description":"šŸšØ May drop, duplicate, misroute, suppress, or wrongly target messages."}],"state":"open","locked":false,"assignees":[],"milestone":null,"comments":1,"created_at":"2026-06-07T01:07:16Z","updated_at":"2026-06-16T12:31:19Z","closed_at":null,"assignee":null,"author_association":"CONTRIBUTOR","issue_field_values":[],"type":null,"active_lock_reason":null,"draft":false,"pull_request":{"url":"https://api.github.com/repos/openclaw/openclaw/pulls/91049","html_url":"https://github.com/openclaw/openclaw/pull/91049","diff_url":"https://github.com/openclaw/openclaw/pull/91049.diff","patch_url":"https://github.com/openclaw/openclaw/pull/91049.patch","merged_at":null},"body":"## Summary\n\n- Preserves terminal `chat.send` ACK status (`ok`, `timeout`, `error`) through clients/helper paths that previously reduced the response to \"has a run id\".\n- Clears, restores, or fails pending/optimistic state when the gateway has already returned a terminal ACK, so clients do not wait for run events that will not arrive.\n- Covers Web queued/direct sends, detached `/btw`, queued `/steer`, `/redirect` via `sessions.steer`, Android chat/mic/Talk flows, native Talk flows, TUI, ACP, and the Skill Workshop forwarded schema.\n- Handles the follow-up TUI detached-send ACK case: terminal `/btw` / `/side` ACKs now clear local BTW tracking, and detached accepted-runId replacements rekey local tracking instead of being misclassified later.\n- Updates Android chat, mic, Swift chat, and TUI normal-send timeout/error handling so terminal `timeout` / `error` ACKs surface failed-send state instead of looking accepted.\n\nTerminal ACKs are part of the `chat.send` contract. If a client treats one as an accepted active run, a message, mic turn, or queued command can get stuck or disappear without failure copy.\n\nClients branch on ACK status instead of only checking `runId`: active statuses keep the existing pending-run flow, while terminal statuses clear pending state, restore retryable queue/draft state where appropriate, or surface failure copy.\n\n- Redesigning the public SDK `Session.send` API shape.\n- Standardizing every cross-client terminal-timeout message/toast policy.\n- Broader lifecycle cleanup races not introduced by this PR.\n\n- Whether each touched consumer preserves `runId` and `status` until the final caller branches.\n- Whether terminal `ok` / `timeout` / `error` branches clear or restore pending state without breaking non-terminal `started` / `in_flight` behavior.\n- Whether Android chat/mic and Swift chat now surface terminal timeout/error as failed-send state.\n\n## Linked context\n\nCloses #91048\n\nRelated #84176\nRelated #84306\n\nNo direct maintainer request; this follows the terminal ACK contract through additional consumers and addresses the Android terminal-timeout feedback.\n\n## Real behavior proof (required for external PRs)\n\n- Behavior or issue addressed:\nTerminal `chat.send` / forwarded ACKs with `status: \"ok\" | \"timeout\" | \"error\"` should not be treated as accepted active runs. Live proof exercises terminal `timeout`; current `9cd2b839bbf4a50882564024fbc3f9c003818e89` preserves the Android/Swift/TUI follow-ups and is clean against latest `main`.\n\n- Real environment tested:\nLocal Linux checkout using repo test runners, Android JVM/Robolectric, and a real local Gateway server + WebSocket RPC harness driving Control UI `sendChatMessage` against a terminal `timeout` ACK. Hosted checks on `603380bfc4fbc1604ac16cb6915837191a36ad53` covered macOS Swift, Android build/test, lint, production/test types, and the real-behavior proof checker. Current head adds CI-red triage plus the iOS SwiftFormat and compile-safe logger fixes.\n\n- Exact steps or command run after this patch:\n\n```bash\nnode scripts/run-vitest.mjs packages/gateway-protocol/src/schema/agents-models-skills.test.ts src/gateway/server-methods/talk.test.ts src/tui/gateway-chat.test.ts src/tui/tui-command-handlers.test.ts src/acp/translator.lifecycle.test.ts ui/src/ui/chat/slash-command-executor.node.test.ts\ncd apps/android && ANDROID_HOME=/opt/android-sdk ANDROID_SDK_ROOT=/opt/android-sdk ./gradlew :app:testPlayDebugUnitTest --tests ai.openclaw.app.gateway.ChatSendAckTest --tests ai.openclaw.app.chat.ChatControllerTerminalAckTest --tests ai.openclaw.app.voice.MicCaptureManagerTest\npnpm tsgo:core && pnpm tsgo:test:ui && pnpm tsgo:test:packages\ngit diff --check\n\n# Current-head conflict-repair validation.\nnode scripts/run-vitest.mjs src/tui/tui-command-handlers.test.ts\nnode scripts/run-vitest.mjs run --config test/vitest/vitest.tui-pty.config.ts\nnode scripts/run-oxlint.mjs src/tui/tui-backend.ts src/tui/gateway-chat.ts src/tui/embedded-backend.ts src/tui/tui-command-handlers.ts src/tui/tui-command-handlers.test.ts\nnode scripts/run-tsgo.mjs -p test/tsconfig/tsconfig.core.test.json --incremental --tsBuildInfoFile .artifacts/tsgo-cache/core-test.tsbuildinfo\ngit diff --check\ngit merge-tree --write-tree upstream/main HEAD\n\n# Temporary uncommitted live proof harness copied to src/gateway/pr91049-live-terminal-ack-proof.test.ts, then removed.\nPR91049_HEAD=c18e743ba4ced535ec5351ba536a542db3d7621f node scripts/run-vitest.mjs run --config test/vitest/vitest.gateway.config.ts src/gateway/pr91049-live-terminal-ack-proof.test.ts\n```\n\nGateway/Control UI live proof still applies because later heads changed Android/Swift/TUI clients only. Current head `9cd2b839bbf4a50882564024fbc3f9c003818e89` is the current conflict-repair head and remains clean against latest `main`; it preserves the normal TUI timeout/error failure copy after `loadHistory()`.\n\n- Evidence after fix (screenshot, recording, terminal capture, console output, redacted runtime log, linked artifact, or copied live output):\n\nFocused suite output from this branch:\n\n```text\nTest Files 9 passed (9)\nTests 237 passed (237)\nAndroid JVM focused ACK suites: BUILD SUCCESSFUL\nType checks: pnpm tsgo:core, tsgo:test:ui, tsgo:test:packages passed\nTouched TypeScript oxlint: Found 0 warnings and 0 errors\n```\n\nCurrent-head local validation on `9cd2b839bbf4a50882564024fbc3f9c003818e89`:\n\n```text\nTUI ACK 66/66; TUI PTY 13/13; focused TUI oxlint 0/0; core-test typecheck exit 0; merge-tree clean.\nCI-red stale-base replays: prompt snapshots current; session-snapshot 8/8; preview-warnings 42/42; doctor-config-flow 39/39.\nLatest-base replays: tooling helper route test passed; extensions test typecheck passed.\nSwift targets: line 1045 uses self.waitForChatCompletion; logger status messages avoid OSLogMessage concatenation.\n```\n\nLive Gateway + Control UI terminal-ACK proof from head `c18e743ba4ced535ec5351ba536a542db3d7621f` for the unchanged Gateway/Control UI path:\n\n```text\nāœ“ gateway-core ../../src/gateway/pr91049-live-terminal-ack-proof.test.ts (1 test)\nāœ“ gateway-server ../../src/gateway/pr91049-live-terminal-ack-proof.test.ts (1 test)\nāœ“ gateway-client ../../src/gateway/pr91049-live-terminal-ack-proof.test.ts (1 test)\nTest Files 3 passed (3)\nTests 3 passed (3)\n```\n\nCopied proof artifact from that run:\n\n```json\n{\"terminalStatus\":\"timeout\",\"gatewayTerminalAck\":{\"runId\":\"00000000-0000-4000-8000-000000091049\",\"status\":\"timeout\"},\"uiResult\":null,\"uiStateAfter\":{\"chatSending\":false,\"chatRunId\":null,\"chatStream\":null,\"lastError\":\"The run ended before the message was accepted.\"}}\n```\n\nHosted PR checks on broad-validation head `603380bfc4fbc1604ac16cb6915837191a36ad53`: `macos-swift`, `android-build-play`, `android-test-play`, `android-test-third-party`, `check-lint`, `Real behavior proof`, and `Auto response` were `SUCCESS` (runs: https://github.com/openclaw/openclaw/actions/runs/27089311492, https://github.com/openclaw/openclaw/actions/runs/27089323348, https://github.com/openclaw/openclaw/actions/runs/27089323350).\n\n- Observed result after fix:\nFocused suites cover protocol schema, gateway Talk, TUI, ACP, slash commands, Android chat, and Android mic. Live Gateway proof shows terminal `timeout` over WebSocket to Control UI `sendChatMessage`; after ACK, pending state is cleared and failure copy is set. TUI tests cover detached `/btw` / `/side`, accepted-runId rekeying, and normal-send timeout/error failure copy after `loadHistory()`. Current-head validation adds latest-base stale-failure replays, clean merge-tree, diff check, focused TUI checks, TUI oxlint, core-test typecheck, and the iOS SwiftFormat/logger fixes.\n\n- What was not tested:\nFull browser/device manual UI testing was not performed. The live proof exercises the Control UI controller source and Gateway WebSocket transport, not a screenshot-driven browser session.\n\n- Proof limitations or environment constraints:\nThe live proof controls the agent reply boundary through the repository Gateway test seam so `chat.abort` can deterministically produce a terminal `timeout` ACK. Gateway transport, idempotency caching, terminal ACK payload, and the Control UI controller cleanup path are real source code. The temporary proof harness was not committed to the feature branch. Swift execution was validated by hosted `macos-swift`, not by the local Linux checkout.\n\n- Before evidence (optional but encouraged):\nFocused RED tests were used during development for Web helper callers, Android chat timeout/error acceptance, Android mic terminal timeout, ACP terminal `ok`, and `/redirect` terminal ACK handling before the fixes were applied.\n\n## Tests and validation\n\n- Focused JS suite listed in the Real behavior proof section.\n- Focused Android JVM/Robolectric ACK suites for `ChatSendAckTest`, `ChatControllerTerminalAckTest`, and `MicCaptureManagerTest`.\n- `pnpm tsgo:core`\n- `pnpm tsgo:test:ui`\n- `pnpm tsgo:test:packages`\n- Touched TypeScript oxlint.\n- `git diff --check`\n- Hosted PR checks listed above on head `603380bfc4fbc1604ac16cb6915837191a36ad53`.\n- Focused TUI ACK suite on `9cd2b839bbf4a50882564024fbc3f9c003818e89`: 66 tests passed, covering normal timeout/error/ok, detached `/btw` / `/side`, rekeying, and timeout feedback after `loadHistory()`.\n- Current-head gates: previous TUI conflict checks plus latest-base CI-red direct replays; iOS line 1045 now uses `self.waitForChatCompletion`; logger status messages no longer concatenate `OSLogMessage`.\n\n- Web direct send, queued send, detached `/btw`, queued `/steer`, and `/redirect` terminal ACK behavior.\n- Android ACK parsing, chat controller terminal ACK handling, and mic terminal ACK handling.\n- Swift chat terminal timeout feedback.\n- Gateway Talk terminal ACK rejection before realtime subscription.\n- ACP terminal `ok` parity with terminal timeout/error.\n- TUI terminal ACK send-result handling, including normal-send timeout/error failure copy, detached `/btw` / `/side` terminal ACK cleanup, and detached accepted-runId rekeying.\n- Skill Workshop forwarded ACK schema accepting the widened terminal statuses.\n\nRED coverage showed terminal ACK helper callers, Android chat, and Android mic handling could still collapse, accept, or discard terminal statuses after the initial propagation work. `/redirect` also treated terminal forwarded ACKs as a successful redirect with a tracked run id.\n\nTests were added or updated for the main status-preserving branches. Swift source changes include test updates; hosted `macos-swift` passed on the final head.\n\n## Risk checklist\n\nYes. Clients now clear, restore, or fail pending state when the gateway returns a terminal ACK instead of waiting for future run events.\n\nNo.\n\nNo.\n\nSession/chat lifecycle state: clearing pending or optimistic state too early could hide a message, while not clearing it leaves clients stuck.\n\nThe patch preserves non-terminal `started` / `in_flight` behavior, keeps `runId` remapping where needed, restores retryable draft/queue state on terminal failure, and adds regression coverage for the changed branches.\n\n## Current review state\n\nReview the status-preserving ACK branches, Android/Swift timeout follow-up, TUI detached ACK cleanup, normal TUI timeout feedback, and current head `9cd2b839bbf4a50882564024fbc3f9c003818e89`.\n\n- Hosted PR checks will report post-push status; no `9cd2b83` hosted run has executed yet.\n- The real-behavior proof gate and platform checks completed successfully on head `603380bfc4fbc1604ac16cb6915837191a36ad53`; head `e5b96a8c970630618d71b665388c1b924fd07863` adds local focused TUI validation for the detached ACK follow-up.\n- Current head `9cd2b839bbf4a50882564024fbc3f9c003818e89` is merge-tree clean against latest `main`; latest-base stale-failure replays passed; the iOS SwiftFormat/logger targets are fixed.\n- This PR is waiting on final bot/maintainer review for the current head.\n","closed_by":null,"reactions":{"url":"https://api.github.com/repos/openclaw/openclaw/issues/91049/reactions","total_count":2,"+1":1,"-1":0,"laugh":0,"hooray":0,"confused":0,"heart":0,"rocket":0,"eyes":1},"timeline_url":"https://api.github.com/repos/openclaw/openclaw/issues/91049/timeline","performed_via_github_app":null,"state_reason":null}]