feathers-mcp defines 4 tools with basic schemas and descriptions. However, the server has significant gaps in parameter documentation, output schema specification, error handling guidance, and tool composition. Tool names lack action verbs (whoami, hidden_tool, boom, echo_json are not verb_noun patterns). Descriptions are minimal (under 100 chars each) and lack context for when/why to use each tool. Input schemas are present but parameters lack descriptions. Output schemas are not documented. The test tools appear minimal/incomplete for a production framework. Code shows proper TypeScript typings and Zod integration, but MCP-specific quality patterns are underutilized. Average tool score: 52 across 4 tools.
Always throws
Echoes its input back as structured JSON
Should never reach MCP clients
Returns the email of the authenticated user
Tool names do not follow verb_noun pattern. Names like 'whoami', 'hidden_tool', 'boom' lack action verbs (get_, create_, delete_, list_, etc.) that help LLMs infer intent before reading descriptions.
Parameter descriptions are completely absent. The 'delayMs' parameter in whoami and 'value' parameter in echo_json have no descriptions explaining purpose, constraints, or expected format. LLMs cannot infer parameter meaning from names alone.
Output/return schemas are not documented. No visible documentation of what fields each tool returns or their types. Tool descriptions do not explain the response structure. LLMs cannot plan downstream operations without knowing response format.
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 43 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 19 | - | v1 |
Tool descriptions are minimal and lack context. 'Returns the email of the authenticated user' (whoami), 'Should never reach MCP clients' (hidden_tool), 'Always throws' (boom) do not explain when/why to use each tool or any error handling guidance.
No error handling guidance for tools with known error cases (boom always throws, whoami requires authentication). Descriptions do not explain recovery steps or classify errors as retryable/user-fixable.
Test tools lack production relevance. whoami, hidden_tool, boom, and echo_json appear to be minimal test fixtures rather than real production tools. This suggests the framework documentation or real examples are not included in the evaluated codebase.
hidden_tool description indicates it should not be exposed to MCP clients, but tool is registered and visible. Either remove it or clarify its intended use (e.g., internal-only flag with documentation).