MCP Health
Run a live health check on MCP servers to distinguish config, auth, and connectivity problems.
Steps
- Check enabled servers — list which MCP servers are configured and enabled
- Test connectivity — attempt a live connection or status check per server
- Classify failures — config missing, auth expired, provider unreachable, or unknown
- Report next fix — the smallest action to resolve each failure
Output
Server: <name> Status: OK | FAIL (<reason>)
Server: <name> Status: OK | FAIL (<reason>)
Action: <next fix for first failure>
Rules
- Never expose secret values or tokens in output
- Never claim MCP health from config file presence alone — test live
- Never invent provider-specific health commands that don't exist locally
- Distinguish config problems from auth problems clearly