All MCP examples
analyze_architecture — TypeScript / express (human-readable)
Request
- Tool:
analyze_architecture - Parameters:
{ "project_path": "." }(or absolute path to cloned express repo) - Optional:
preset: "ai"to run agent_architecture + openclaw_architecture; omit or use default to run configured metrics (e.g. scc only).
What the tool does
Runs Arxo analysis and returns agent architecture scores (reliability, governance, safety, coordination) and OpenClaw architecture scores (config security, skill governance, observability, supply chain), plus per-detector findings with file, line, and recommendation.
Example response (summary)
- agent_architecture.overall_agent_health — Composite 0–1 score. Higher is better. For a non-agent repo like express, this may be N/A or low if no agent patterns are detected.
- agent_architecture.agent_reliability_score — Reliability axis (e.g. loop guards, timeouts).
- findings — List of issues with
rule_id,severity,evidence(file, line, reason),recommendation. Address high-severity first. - openclaw_architecture — Only present if the project has
openclaw.json(or.openclaw/openclaw.json). For express, typically absent or empty.
What to do next
- Check
overall_agent_health(and axis scores) for a quick health signal. - Triage
findingsby severity; userecommendationandevidenceto fix issues. - For OpenClaw projects, review config and skill governance findings.
Full request/response
See analyze_architecture.json for the raw request and response payload (when available).