Two tools with basic schemas and descriptions, but significant gaps in parameter documentation and output structure. code_collector accepts union types (string|list[string]) without clear guidance on behavior differences. project_structure_reporter lacks parameter descriptions entirely. Neither tool documents return structure or error recovery paths. Descriptions are present but generic (10-20 words). No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite both being read-only operations. Error handling returns plain strings without actionable guidance.

• Add parameter descriptions to project_structure_reporter: 'output_filename: Name of the markdown report file to write to .neurolora/ directory (default: PROJECT_STRUCTURE_REPORT.md)' and 'ignore_patterns: List of glob patterns to exclude from analysis (e.g., ["*.pyc", "__pycache__", "node_modules"])'.
• Document code_collector input_path behavior: 'input_path: Single file path (string) or list of paths (list[string]). Directories are recursively scanned for code files. Default "." scans project root.'
• Return structured JSON objects instead of plain strings. E.g., code_collector should return {"status": "success", "output_file": "/path/to/file.md", "file_count": 42, "total_lines": 5000} to enable downstream tool chaining.
• Add error classification and recovery hints: 'Error: No files matched input_path. Verify path exists and is readable. Try code_collector with a broader path like "." to scan the entire project.'
• Add readOnlyHint: true to both tool definitions in FastMCP registration to signal these are safe, non-destructive operations.

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

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