Server exposes 7 tools with basic descriptions and input schemas. Tool names follow verb_noun convention (search_*, list_*, remember, recall, get_*). Descriptions are present but brief (avg ~60 chars, below the 194-char baseline). Input schemas are visible and typed, but parameter descriptions are minimal or missing context. No output schemas documented. Error handling and recovery guidance absent. The 'remember' tool is WRITE-capable but lacks confirmation/dry-run pattern. Overall: functional definitions with gaps in LLM-optimization and error guidance.

• Document output schemas for all 7 tools. Include field names, types, and whether results are paginated. E.g., search_hex_packages should return [{name, description, url, downloads}] with total_count.
• Expand tool descriptions to 100 - 200 chars. Add WHEN to use (e.g., 'Use search_docs for semantic package documentation; use search_hex_packages for package discovery by name'). Include return value hints.
• Add min/max bounds to numeric parameters. E.g., limit: 1 - 100 (default 20). Document cost of refresh=true in search_docs (re-ingestion time).
• Implement confirmation pattern for 'remember' tool. Return a preview of the fact to be stored and require explicit confirmation before writing to knowledge base.
• Add error recovery guidance. E.g., 'If search_docs returns no results, try search_hex_packages to verify the package exists, then call search_docs with refresh=true.'

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

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