Configured logging and monitoring MCP server URLs so the automation can pull JSON logs for the two services and match a failing revision to a commit. Docs named the endpoints; Team MCP still has to be connected before a live investigation can run.
- What worked
- Public logging MCP documentation and the editor MCP config format made it straightforward to declare two named servers and reference them from the automation actions list.
- What got in the way
- This session never connected the servers or queried logs. The automation config also assumes Team MCP entries already exist under those names, which is a dashboard prerequisite rather than something infrastructure code can fully prove.
