Single tool 'baadiff_scan' has a basic description (89 chars) but critically lacks parameter descriptions. The input schema defines 'target' as a string with a description, but the tool description itself is generic and does not explain WHEN to use it, WHAT it returns structurally, or HOW to interpret results. No output schema is documented. The tool name 'baadiff_scan' is verb-noun compliant, but the overall definition is sparse and would not guide an LLM effectively on result interpretation or error recovery.

• Add explicit output schema to tool description: 'Returns a JSON object with keys: score (0-100 integer), findings (array of {check_id, title, severity, status, message, file, line}), summary (string).'
• Expand parameter description for 'target': 'Absolute or relative filesystem path to a Git repository root or infrastructure manifest file (Terraform, CloudFormation, Kubernetes YAML). Max 2GB. Returns error if path does not exist or is not readable.'
• Document error cases: 'If path is invalid, returns error with message. If no supported files found, returns empty findings array with score 0. If permission denied, returns error: request read access to target directory.'
• Add result limit guidance: 'Findings are capped at 1000 per scan. For large repos, consider scanning subdirectories separately.'
• Include example output structure in description or as a separate schema reference so LLM understands the JSON shape before calling the tool.