Two tools with clear, concise descriptions and basic parameter schemas. search_docs has proper input schema with query (string) and k (integer, default=3). answer_question has simpler schema with just query. Both descriptions are action-oriented and under 100 chars, meeting baseline. However, parameter descriptions are minimal (query lacks detail on format/length), output schemas are undocumented (responses are free-text strings, not structured objects), and no error handling guidance is provided. No tool annotations (readOnlyHint, idempotentHint) despite both being read-only. Naming is clear (verb_noun pattern) but generic. Missing: pagination for search_docs, structured output format, error recovery hints.

• Document output schemas explicitly. search_docs should return {results: [{doc_id: string, score: number, snippet: string}], total: number}. answer_question should return {text: string, answered: boolean, citations: [{doc_id, title, snippet}]}.
• Add parameter constraints: query (1-500 chars, required), k (1-50, default 3). Include in descriptions: 'Search query (1-500 characters, required). Supports natural language and keywords.'
• Add error recovery hints. search_docs: 'If no results, try broader keywords or check your access permissions.' answer_question: 'If refused, the query may not match accessible documents, try rephrasing or search_docs() first.'
• Implement pagination for search_docs: add offset (default 0) and limit (default 10, max 50) parameters. Return {results: [...], total: number, has_more: boolean}.
• Add tool annotations in FastMCP registration: @server.tool(readOnlyHint=True, idempotentHint=True) for both tools to signal safety to agents.