Two tools with clear, security-focused descriptions and proper schemas. scan_site has well-defined enum for mode parameter and URL format constraint. scanner_policy is a simple read-only introspection tool. Both tools have descriptions (194 and 115 chars respectively, within baseline 34-392 range). Input schemas are present with type definitions. However, output schemas are not documented in the source code provided, and parameter descriptions lack detail on expected behavior, error cases, and recovery guidance. No evidence of pagination, result limits, or structured output documentation. Error handling strategy is not visible.

• Document output schema for scan_site: specify SARIF structure, fields returned (findings array, tool metadata, etc.), and example result shape. Include total finding count and any pagination/truncation logic.
• Expand scan_site mode parameter description: 'passive (default): read-only checks only, safe for any URL. authorized: adds read-only exposure probes (e.g., header injection tests); requires explicit permission from site owner. Unauthorized use may violate CFAA or local laws.'
• Add error handling guidance to scan_site: 'If URL is unreachable, returns timeout error with retry guidance. If URL is invalid, returns validation error with format requirements. If scan is rate-limited, returns 429 with backoff hint.'
• Clarify scanner_policy description: 'Returns a JSON object documenting this tool's safety boundaries: max scan duration, URL allowlist/blocklist, authorized mode restrictions, and agent context protections. Call this first to understand what this tool can and cannot do.'
• Document result limits: 'scan_site returns up to 100 findings per scan. If more findings exist, response includes a truncated flag and guidance to run targeted scans on specific categories.'