Both tools have minimal descriptions (Chinese text, ~20-30 chars each) that fail to explain WHAT the tool does, WHEN to use it, or prerequisites. Schemas are present but lack parameter descriptions entirely. query_documents accepts only 'question' with no type info visible; vectorize_directory accepts 'directory_path' with no description. No output schemas documented. Error handling is absent, no guidance on recovery or retry logic. Tool names are acceptable (verb_noun pattern) but descriptions are critically deficient. This server falls well below production baseline.

• Rewrite tool descriptions in English, 100-200 characters each. Example for query_documents: 'Search the vector database for document fragments matching a user question. Returns up to 5 most relevant chunks with source metadata. Use this to answer questions about loaded documents.'
• Add descriptions to all parameters. Example: 'question (string, required): The user's question or search query. Can be a natural language phrase or keyword search. Max 1000 characters.'
• Document output schemas explicitly. For query_documents, specify: returns {results: [{source: string, content: string, score: float}], total_found: int}. For vectorize_directory, return {status: 'success'|'error', documents_loaded: int, chunks_created: int, error_message?: string}.
• Add error handling with recovery guidance. Example: 'If directory_path does not exist, return {status: "error", message: "Path not found: /invalid/path. Verify the path exists and is readable.", retryable: true}.'
• Add pagination to query_documents: accept optional 'limit' (1-50, default 5) and 'offset' (default 0) parameters. Return {results: [...], total_found: int, has_more: bool} to enable multi-turn retrieval.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry