Stitch MCP CLI helper. Automates Google Cloud authentication. Generates MCP config for your client. Sets up a proxy server.
Stitch MCP exposes only ONE tool: 'list_tools'. This tool has a reasonable name and basic description, but the schema is empty (no parameters), and critically, there is NO documented output schema. The tool serves a meta/discovery purpose rather than providing domain-specific functionality. The codebase shows this is primarily a CLI helper and proxy wrapper for Google Cloud authentication and Stitch SDK integration, not a feature-rich MCP server. The single tool definition lacks parameter descriptions, output documentation, and error handling guidance. Per the hard scoring rules, tools with missing schemas or undocumented outputs cannot score above 40 in schema. The overall definition quality is severely limited by having only one meta-tool with no real domain functionality.
List all available tools with their descriptions and schemas.
No output schema documented for list_tools. The tool returns a list of tools with descriptions and schemas, but the response structure is undefined. LLMs cannot plan downstream actions without knowing the return shape.
Single meta-tool only. The server exposes only 'list_tools', which is a discovery/introspection tool, not domain functionality. There are no actual tools that perform meaningful operations (e.g., managing resources, calling external APIs). The server is primarily a CLI helper and proxy wrapper, not a true MCP server with operational capabilities.
Empty parameter schema. The 'list_tools' tool has an input schema of {"type":"object","properties":{}} with no properties defined. While this correctly indicates 'no parameters required', there is no description of the tool's behavior, preconditions, or the structure of the returned tool list.
Inferred effective spec: 2025-06-18+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 44 | 2025-06-18+ | v2 |
| 2026-03-09 | D | 51 | - | v1 |
Minimal tool description. The description 'List all available tools with their descriptions and schemas' is only 60 characters. While present, it lacks guidance on when to call this tool, what structure is returned, or how the LLM should interpret the output. Descriptions should be 10 - 1024 chars, ideally 50 - 200 for clarity.
No error handling or recovery guidance. If 'list_tools' fails (network error, permission issue, malformed response), there is no documented error message, error codes, or recovery steps for the LLM to follow. Error responses should guide the agent on what to do next.