Server has 6 tools with explicit Zod schemas and descriptions. All tools are properly registered with inputSchema and descriptions. However, descriptions are generic (avg ~60 chars, well below the 194-char baseline for A+ tools), parameter descriptions lack context about constraints and use cases, and output schemas are not documented. Error handling returns JSON with isError flag but lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint) despite clear risk classifications. Naming is verb-first and clear, but descriptions do not explain WHEN to use each tool or dependencies between them.

• Expand tool descriptions to 100-200 chars. Include WHEN to use (e.g., 'Call this after ingest_document to verify content was indexed'), prerequisites, and how it differs from similar tools. Example: 'Search indexed documents by semantic similarity. Call after ingesting documents. Returns top_k results ranked by relevance score. Use instead of list_documents when you need to find specific content.'
• Add tool annotations to schema: set readOnlyHint=true for search/list/get tools, destructiveHint=true for delete_document, and idempotentHint=true for search/list (safe to retry). This guides agent planning without changing behavior.
• Document output schemas in tool descriptions or as a separate schema field. Example: 'Returns {results: [{id, title, score, chunk_text}], total_count}' so LLMs know what fields to expect and can chain calls.
• Enhance error messages with recovery hints. E.g., 'Document not found. Call list_documents() to see available IDs, or search_knowledge_base() to find by content.'
• Add parameter descriptions for metadata and top_k explaining valid structures, typical ranges, and use cases. E.g., 'top_k: Number of results (1-20, default 5). Higher values return more candidates but slower; use 5-10 for most queries.'