Five read-only tools with complete JSON schemas and reasonable descriptions (avg 95 chars). All tools have proper input schemas with type constraints and descriptions. Naming follows verb_noun pattern (read_, search_, list_, get_, route_). However, descriptions are generic and lack LLM-optimized guidance on WHEN to use each tool vs alternatives. No output schemas documented. Error handling is minimal, tools return plain text rather than structured recovery guidance. No parameter descriptions for optional fields (e.g., context_lines, case_sensitive). Missing idempotency hints and chaining metadata.

• Document output schemas for all tools. E.g., search_tools_md should return {matches: [{line: number, block: string}]} or similar structured format, not plain text.
• Enhance descriptions with selection guidance. E.g., 'Use search_tools_md to find tools by keyword. Use list_tools_md_sections to browse all available sections. Use get_tools_md_section when you know the exact heading name.'
• Add descriptions to all optional parameters. E.g., 'context_lines (default 2): number of lines before/after each match for context. Increase for longer excerpts, decrease for brevity.'
• Implement structured error responses with recovery hints. E.g., instead of 'No matches for: X', return {error: 'no_matches', query: 'X', suggestion: 'Try search_tools_md with fewer keywords or check spelling'}.
• Clarify route_task_with_tools_md output format in description. E.g., 'Returns a ranked list of relevant sections and search hits that match the task, ordered by relevance. Use this to guide the agent to the right tools.'
• Add idempotency hints to tool descriptions. All five tools are read-only and safe to retry, state this explicitly so agents know they can safely retry on transient failures.