PolyMCP Operator

Overview

Run structured incident triage for PolyMCP systems and deliver minimal, verifiable remediation.

Quick Triage

Start with this baseline:

polymcp server list --json
polymcp test server http://localhost:8000/mcp
polymcp test tool http://localhost:8000/mcp add --params '{"a":2,"b":3}'

For stdio incidents:

polymcp test stdio npx @playwright/mcp@latest

Incident Flow

1. Scope impact.

• Identify affected server(s), transport (http/stdio), tool(s), and user-visible failure.

2. Check registry.

• Inspect local and global polymcp_registry.json.
• Verify server URL and stdio command/args/env.

3. Reproduce minimally.

• Run server test first.
• Run one failing tool invocation with exact payload.
• Capture raw command outputs.

4. Apply smallest safe fix.

• Prefer config or targeted code fix.
• Avoid broad refactor during active incident.

5. Verify and close.

• Re-run same commands.
• Summarize root cause, fix, evidence, and rollback.

Failure Classes

• Connectivity: endpoint not reachable or wrong route.
• Contract: payload schema mismatch or tool name mismatch.
• Runtime: server exception or dependency error.
• Auth: missing/invalid secret or header.
• Registry: stale/incorrect server entry.

Operating Rules

• Tie every conclusion to command evidence.
• Make smallest reversible change first.
• Keep rollback notes explicit.
• Treat missing secret/token as config issue first.
• Record any residual risk after fix.