Single tool 'aegis_scan' has a well-structured JSON Schema with proper types and required fields. Description is comprehensive (280+ chars) and explains the tool's purpose, supported frameworks, and output types. However, parameter descriptions are minimal: 'target' lacks guidance on path format/validation, and 'format' enum is present but lacks explanation of when to use each. No output schema documented, LLMs cannot predict response structure. Error handling is basic (returns text errors, no recovery guidance). Tool does one thing well (audit), but lacks production-grade parameter validation and output documentation.

• Document the output schema: specify whether response is a JSON object with fields like {permission_matrix: {...}, findings: [...], risk_score: number} or markdown text. Include field types and nesting.
• Enhance 'target' parameter description: 'Filesystem path to the agent project root (absolute or relative). Must be a readable directory containing agent config files (mcp.json, pyproject.toml, package.json, etc.). Max 4096 chars.'
• Add descriptions to format enum values: 'json: Structured output for programmatic parsing; markdown: Human-readable report for display or documentation.'
• Implement structured error responses with recovery hints: 'Target not found at /path/to/project. Verify the path exists and is readable. Supported formats: mcp.json, pyproject.toml, package.json.'
• Add tool annotations (readOnlyHint: true) to signal that aegis_scan is safe to call repeatedly without side effects.