Monorepo of MCP (Model Context Protocol) tools, adapters, middleware, and frameworks. Includes core MVA framework, cloud adapters (AWS, Vercel, Cloudflare), authentication (API Key, JWT, OAuth), code generators (OpenAPI, Prisma), and integration connectors (n8n, A2A, Swarm).
MCPFusion exposes only 2 tools with minimal schema documentation and sparse descriptions. The 'validate' tool has a basic input schema but lacks depth in its description (26 chars); the 'status' tool has NO input parameters defined and a vague description (48 chars). Neither tool description explains when to use them, what they return, or their preconditions. Output schemas are not documented anywhere in the visible source. The server appears to be an authentication middleware framework rather than a full-featured MCP server, and the tool definitions reflect this, they are skeletal and insufficient for production LLM use. No evidence of error handling guidance, parameter validation messaging, or multi-step composition patterns. Both tools are READ_ONLY, which is good for safety, but this alone does not compensate for missing schemas and descriptions.
Check API key authentication status from context
Validate an API key
Output schemas are not documented. Neither 'validate' nor 'status' tool describes what they return to the LLM. Without return type documentation, LLMs cannot plan downstream operations or extract relevant data.
Tool descriptions are too brief and lack context. 'Validate an API key' (26 chars) does not explain when to call it vs 'status', what validation rules apply, or what success/failure looks like. 'Check API key authentication status from context' (48 chars) is generic and does not guide LLM selection.
The 'status' tool has NO input schema. No parameters are defined, which means either the tool takes no input (valid but undocumented) or the schema is missing from the source. Either way, LLMs cannot know what to pass.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 38 | 2026-07-28+ | v2 |
Parameter descriptions are missing or trivial. The 'validate' tool's 'key' parameter description is 'The API key to validate', this does not explain format, length, character restrictions, or what constitutes a valid key. LLMs will guess and pass invalid values.
No error handling guidance. If validation fails, what should the LLM do? Retry? Ask the user for a different key? The tools provide no recovery hints, leaving agents stranded on errors.
Tool composition and chaining is unclear. Does 'status' return the same validated key data as 'validate'? Can they be called in sequence? The lack of documented output fields breaks the tool-chain pattern.