MCP server that exposes workspace context to any MCP-compatible AI agent. Agents get service topology, health status, KB docs, and blast radius analysis without needing Orcha-specific agent skills.
Orcha MCP server defines 5 read-only tools with clear verb-noun naming (get_*, search_*). All tools have descriptions (avg 95 chars) and documented input schemas with type info. However, output schemas are not visible in source code, parameter descriptions lack detail (e.g., no format constraints, ranges, or dependency hints), and error handling guidance is absent. No tool annotations (readOnlyHint, etc.) despite all being read-only. Schemas are present but minimal, parameters lack descriptions beyond the name. This is a solid foundation but falls short of production-grade LLM-optimized tooling.
Analyze what services, probes, and flows would be affected if a service changes or goes down
Get the resolved configuration for a service, optionally with a specific profile applied
Get the dependency-resolved startup order for a preset or service
Get a high-level summary of the workspace: services, presets, defaults, and config location
Search knowledge base documents across all services. Returns matching file paths and snippets.
Output schemas not documented in source. LLMs cannot plan downstream tool calls or extract required fields without knowing response structure.
Parameter descriptions are minimal or absent. 'serviceId' and 'profile' lack format hints, constraints, or dependency documentation. LLMs cannot validate inputs or understand when to use optional params.
No tool annotations despite all tools being read-only. Missing readOnlyHint would help LLMs understand these are safe to call without side effects.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 58 | 2026-07-28+ | v2 |
No error handling guidance. Tools do not document what happens on invalid input (e.g., unknown serviceId, malformed query) or how LLMs should recover.
search_kb accepts optional serviceId but no pagination (limit, offset, cursor). Large knowledge bases could return unbounded results, exhausting context window.