Claw Background Jobs

Claw Background Jobs Skill

When to Use

Use x agent run for tasks estimated >2 minutes or that would block the chat:

• Multi-step tasks
• Data analysis
• Research
• Anything expected to take >2 minutes

When NOT to Use

• Quick Q&A or simple commands (reply directly)
• Tasks needing back-and-forth clarification (stay in chat)

Commands

• x agent run --job-id "<id>" --max-iterations <n> "<task>": Create and start an async job.
• x agent job status --job-id "<id>" --llms: Check job state and progress.
  • After checking status, go to the jobdir directory shown in the output.
  • Read key files inside jobdir to get concrete details:
    • DESC.md or AGENTS.md — task description and context
    • status.yml — detailed runtime state
    • *.response.md files — iteration logs with actual results
  • Pay attention to ai.recommendation and progress.todo in the YAML output.
• x agent job ls --active --llms: List active jobs.
• x agent job stop --job-id "<id>": Stop the background process.

Job ID Convention

Use <im>-<chatid>-<brief> for traceability, e.g., weixin-user123-loganalysis.

Critical Constraints

• x agent run executes in a fresh environment with NO chat memory. The <task> must be self-contained: clear goal, required tools, expected output.
• The job does NOT notify the user automatically. Claw checks status via heartbeat and uses x claw agentrequest to report results.