Three tools with clear verb-noun naming (docs_list_sources, docs_search, docs_get_chunk) and reasonable descriptions (50-100 chars). All tools have input schemas with type definitions and parameter descriptions. However, output schemas are not documented in the visible code, parameter descriptions lack detail about format/constraints, and no error handling guidance is provided. Tools follow single-responsibility principle but lack pagination guidance for list_sources despite accepting a limit parameter.

• Document output schemas for all three tools. Example for docs_search: 'Returns array of {chunk_id: string, text: string, source: string, score: number, offset: number}' so LLMs know what fields to extract.
• Enhance parameter descriptions with constraints. For 'query': 'Search query text (1-500 chars, supports AND/OR/NOT operators)'. For 'chunk_id': 'The chunk identifier (format: source_name:chunk_number, e.g. readme.md:0).'
• Add error recovery guidance to tool descriptions. E.g. docs_search: 'If no results found, try broader query terms or call docs_list_sources to discover available sources first.'
• Clarify pagination for docs_list_sources. State: 'Returns up to [limit] sources sorted by [field]. To fetch all sources, call repeatedly with offset parameter (if supported) or iterate until returned count < limit.'
• Explain tool selection criteria. Add to docs_search: 'Use this for semantic/keyword search across indexed documents. Use docs_list_sources to discover available files before searching.'

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

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