| name | runtime-operator |
| description | Operate and regression-test a local ccbot runtime through CLI, including worker control, memory commands, schedule commands, and observability checks. |
| metadata | {"ccbot":{"emoji":"🛠️"}} |
Runtime Operator
Use this skill when the goal is to run, rehearse, or validate the agent runtime locally.
Preferred path
Start with CLI before Feishu or other bot channels.
uv run ccbot run --config ~/.ccbot/config.json --channel cli
Minimum regression set
Check these in order:
/help
/workers
/memory show
/schedule list
- a normal user message
- a task likely to dispatch workers
- a recurring schedule request, then
/schedule run <job_id>
What to watch
- startup log prints model / workspace / LangSmith state
- control commands should return immediately without hitting the model
- worker results should come back through the channel path
- scheduler jobs should persist under
workspace/.ccbot/schedules/
- failures should expose
sdk stderr and LangSmith traces
If runtime fails
Collect:
- triggering message
- last 30-50 log lines
- any
[sdk:...] STDERR | ...
- LangSmith trace link or last tool name