This server declares a single tool 'dobby_mcp_tools' with only a high-level description ('90+ tools for task management...') but provides NO visible tool definitions, schemas, parameters, or structured output documentation in the source code. The tool is instantiated via MCPToolset wrapping an external npm package (@dobbyai/mcp-external), meaning tool definitions are not inspectable here. Per the hard scoring rule: 'If you cannot see the actual tool definition in the source (only inferred): cap that tool's overall at 50.' Since the tool is entirely opaque and delegated to an external binary, and no schemas or parameter documentation are visible, this scores as a 5, barely functional declaration with zero verifiable structure.

• Expose the actual tool definitions from @dobbyai/mcp-external in the source or documentation. List each of the 90+ tools with name, description, parameters, and output schema.
• For each tool, provide a verb_noun name (e.g., create_task, search_knowledge_base, approve_workflow) and a 50 - 200 character description explaining WHEN to use it.
• Document parameter types, constraints (enums, ranges), and descriptions. E.g., 'status: enum(open|in_progress|closed), the task status to filter by.'
• Add error handling guidance: 'If the API returns 401, check DOBBY_API_KEY. If 404, the resource may have been deleted, try listing available resources.'
• Create a schema registry or OpenAPI spec for the 90+ tools so LLMs can reason about them without guessing.