Two tools with reasonable naming and descriptions, but significant gaps in schema documentation and parameter clarity. search_files has a detailed description (450+ chars) explaining preview behavior and path_prefix semantics, which is excellent. However, neither tool's input schema is fully visible in the provided source code, only parameter names and descriptions are shown, not formal JSON Schema with types. get_server_info lacks any visible input schema. Output schemas are not documented. Error handling is minimal, no recovery guidance or categorization. The server uses fastmcp framework but does not expose tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite both tools being read-only operations.

• Expose formal JSON Schema for all tool inputs in the MCP tool registration. For search_files, explicitly define query as type string (required), path_prefix as type string (optional), with minLength/maxLength constraints if applicable.
• Document output schema for search_files. Specify that it returns a string with a header section (pipeline summary + score interpretation) followed by numbered results in 'path, score' format. If preview_chars is enabled, note that file content is appended.
• Expand get_server_info description to 100+ chars explaining what 'statistics' are returned (e.g., 'Returns database name, indexed file count, embedding model, and reranker status'). Add examples of when to call it (e.g., 'Call before search to verify the codebase is indexed').
• Add error handling with recovery guidance. When CODEBASE_PATH is missing, return 'CODEBASE_PATH not set. Configure it in your MCP client settings or environment.' When search returns no results, suggest 'No files found. Try a broader query or check path_prefix is correct.'
• Annotate tools with readOnlyHint=true in fastmcp registration to signal to clients that these operations are safe to retry and cache.

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

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