Repository reference
Troubleshooting
Diagnose installation, control-plane, MCP, managed cloud, Docker, Kubernetes, readiness, audit, and incident failures from the first broken layer.
Troubleshooting must preserve request IDs while redacting credentials and payloads. Never weaken readiness, egress, demo, or isolation controls merely to make a check pass.
Use this reference when
- A frozen install, local port, health/readiness, authentication, membership, or endpoint validation check fails.
- A session, tool discovery/call, upstream network, stdio, Wrangler, Compose, or Helm path behaves unexpectedly.
- Collecting a minimal diagnostics bundle or responding to audit-integrity and sensitive-data concerns.
What it covers
Workspace and control plane
Version and install checks, port conflicts, health versus readiness, authentication, membership, and endpoint validation.
Gateway and deployments
Sessions, list/call parity, schemas, remote egress, stdio, managed cloud bootstrap, Compose, Kubernetes, DNS, and probes.
Diagnostics and help
Audit continuity, sensitive-data response, safe reproduction details, request correlation, and private security reporting.
Read the full repository reference
This page is a discovery map, not a shortened replacement for the maintained source. The complete reference includes exact tables, commands, boundaries, and cross-links in docs/troubleshooting.md.
Open the complete Markdown reference ↗