Single tool 'index_repository' has a basic schema with two parameters (repo_path, mode) but lacks critical quality markers. Description is minimal (47 chars) and does not explain WHEN to use the tool, what it returns, or consequences of indexing. Parameters lack detailed descriptions, 'mode' is vague ('e.g., full') with no enum constraint or explanation of valid values. No output schema documented. No error handling guidance. Tool performs a WRITE operation (builds a codebase graph) but description does not warn of side effects or idempotency. Schema is present but incomplete: parameters have types but descriptions are too terse to guide LLM selection or parameter construction.

• Expand tool description to 100-200 chars: explain that indexing builds a semantic graph of the repository, state when to call it (e.g., 'Call once per repository to enable code search and analysis'), and describe the output (e.g., 'Returns indexing status, file count, and graph statistics').
• Add enum constraint to 'mode' parameter with explicit valid values (e.g., 'full', 'incremental', 'fast') and describe what each mode does (speed vs completeness tradeoff).
• Document the output schema: return object with fields like 'status' (success|error), 'files_indexed' (number), 'graph_size' (bytes), 'duration_ms' (number), and 'error_message' (string if failed).
• Add error handling guidance: 'If indexing fails due to invalid repo_path, verify the path exists and is readable. If it times out, try mode=fast for large repos. Indexing is idempotent, safe to retry.'
• Clarify 'repo_path' parameter: 'Absolute or relative path to the repository root. Must be a valid directory containing a .git folder or source files. Example: /home/user/myproject or ./src'

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

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