MathLab/hooks/task-board-realtime/HOOK.md

---

name: task-board-realtime description: "Sync OpenClaw conversations to the task board in real time" homepage: https://docs.openclaw.ai/automation/hooks metadata: {

"openclaw":
  {
    "events": ["message:received", "message:sent", "agent:bootstrap", "command:new", "command:reset", "command:stop"],
    "requires": { "config": ["workspace.dir"] },
  },

}

Task Board Realtime Hook

Automatically writes OpenClaw conversation activity into the local task board.

Behavior

• message:received
  • Creates (or reuses) one root node for the active conversation.
  • Adds each incoming message as a child node.
• agent:bootstrap
  • Fallback path for direct openclaw agent ... turns.
  • Scans recent user messages in the session transcript and appends all unseen turns.
• message:sent
  • Runs a post-response reconciliation pass for channels where direct incoming hooks are not emitted.
  • Resolves session id from sessions.json when needed, then backfills unseen user turns from transcript.
• command:new and command:reset
  • Closes the current root node as completed.
  • Next message starts a new root node.
• command:stop
  • End-of-turn reconciliation for CLI-style runs.
  • Backfills unseen user messages from transcript so the latest turn is not delayed.

Config

Configured under:

hooks.internal.entries.task-board-realtime

Supported fields:

• enabled (boolean, default true)
• apiBaseUrl (string, default http://127.0.0.1:3001)
• adminToken (string, default dev-task-board-token)
• stateFile (string, default <workspace>/task-board/.sync-state.json)
• ignoreSlashCommands (boolean, default true)
• maxRememberedMessages (number, default 3000)