By name: monitor-ci-results description: Monitor remote CI results for a PR and autonomously trigger log analysis upon failures
When the user requests to monitor remote CI results or watch a pull request, invoke scripts/monitor_remote_ci.py <pr_number> <your_conversation_id>.
This long-running monitoring service runs in the background and continuously polls both GitHub PR checks and Buildkite workflow executions.
✨ Autonomous Failure Orchestration
When any CI job completes with errors or returns a non-zero exit code:
- It automatically downloads the raw CI log file to
ci_logs/. - It launches an independent background analyzer script (
analyze_ci_failure.py). - It authors a beautifully structured Markdown suggested plan for how to fix the failure.
- It natively dispatches a high-priority notification message back to your active agent conversation (containing the downloaded log path and fix plan) using
agentapi send-message!
Example Invocation
./scripts/monitor_remote_ci.py 3812 "0be435bd-96aa-4e1b-9c6f-727b31e80fa0" &