Single tool 'query-knowledge-base' has critical definition gaps. Tool description is completely missing (null). All three parameters lack descriptions (null values). Input schema is present with Zod validation (query: string, topK: number 1-10 optional, rerank: boolean optional), but parameter descriptions are absent. Output schema is undocumented, response structure inferred from code (array of objects with content, similarity, score, metadata fields). No error handling guidance. Naming is acceptable (verb_noun pattern) but insufficient to overcome missing descriptions.

• Add a tool description (50-200 chars): 'Query a Ragdoll AI knowledge base to retrieve relevant documents. Returns ranked results with similarity scores. Useful for retrieval-augmented generation (RAG) workflows.'
• Add parameter descriptions: query='The search query string (required)', topK='Number of results to return (1-10, default 5)', rerank='Whether to apply semantic reranking to results (default false)'
• Document the output schema in the tool description or as a structured comment: 'Returns array of {content: string, similarity?: number, score?: number, metadata: object}'
• Add error handling: catch fetch errors and return actionable messages like 'Knowledge base query failed: check RAGDOLL_KNOWLEDGE_BASE_ID and RAGDOLL_API_KEY are set correctly' or 'API timeout, retry with smaller topK'
• Add a README or inline comment documenting required environment variables and setup steps so users know how to configure the server