Critical gaps across all dimensions. Tool definitions lack input schemas (cannot verify parameter types/constraints), descriptions are minimal (10-50 chars, well below 50-200 char baseline for LLM optimization), and no output schemas are documented. Parameters present but untyped. No error handling guidance. Tool names are reasonable (verb_noun pattern) but insufficient to compensate for missing schemas and descriptions. Source code shows tool registration in src/main.c and src/ui/http_server.c but actual schema definitions are not visible in provided excerpts.

• Add complete JSON Schema definitions for all tool inputs. Include type, description, and constraints (enum, minLength, maxLength, pattern) for every parameter. Example: 'mode' should be enum: ['full', 'incremental'] with description explaining when each is appropriate.
• Expand tool descriptions to 80-150 characters. Include WHAT (e.g. 'Searches the code dependency graph'), WHEN (e.g. 'Use this to find all callers of a function'), and WHAT IT RETURNS (e.g. 'Returns matching symbols with file paths and line numbers'). Distinguish search_graph from query_graph in descriptions.
• Document output schemas for all tools. Specify returned fields, types, and whether results are paginated. Example: search_graph returns [{symbol_name: string, file_path: string, line_number: integer, context: string}], paginated with limit/offset.
• Add parameter constraints and examples in descriptions. For 'query' param: 'Free-form search string (1-500 chars). Examples: "function_name", "class MyClass", "import os". Supports regex if prefixed with /.'
• Add error handling guidance to tool descriptions. Example for index_repository: 'On failure, returns error code (invalid_path, permission_denied, disk_full) with recovery hint. Retry is safe for transient errors; ask user to check path/permissions for permanent errors.'

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

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