openclaw-debugging

What the skill teaches

A skill OpenClaw maintainers use to debug the OpenClaw agent itself: model and provider calls, the tool surface, code mode, streaming and behaviour in its Crabbox test environment. The agent names the boundary it suspects, from config and tool construction to provider payloads, SSE streams, transcript replay or packaging, turns on the narrowest logging that would prove it, reproduces with the same provider, model and config, compares configured and actual state, patches the root cause and reruns the exact failing probe. It lists OpenClaw's debug environment flags for model transport, payloads and SSE events, and how to fetch stored sessions and transcripts as evidence.

Capabilities

Details