The 'scan' tool has a description (Image geolocation toolkit...) but the input schema is severely incomplete. The schema shows only two parameters (image_data, image_url) with minimal type info and no descriptions for the parameters themselves. The tool description is 73 chars, which meets the 10-1024 baseline, but parameter descriptions are absent entirely. No output schema is documented. The implementation delegates to cognis_core.mcp.build_mcp_server(), which is not visible in the provided source, making it impossible to verify actual schema registration. Per the hard rules, inferred tool definitions cap at 50; combined with missing parameter descriptions and no visible output schema, this scores low.

• Add explicit parameter descriptions to the schema: 'image_data: Raw image file bytes (JPEG, PNG, GIF, WebP). Max 50 MB. Required.' and 'image_url: Optional URL for reverse-image search link generation. Used only if provided; does not fetch the image.'
• Document the output schema with all returned fields: e.g., {has_exif: bool, gps: {latitude, longitude, altitude_m}, maps_url: string, reverse_search: {google, bing, yandex}, exif_tags: {...}, ocr_text: string, errors: []}
• Rename tool from 'scan' to 'analyze_image' or 'geolocate_image' to clarify the action and match the demo intent (OSINT EXIF triage, verification).
• Add error handling examples in the description: 'Returns error if image_data is not a valid image format. Returns partial results if EXIF is missing but OCR succeeds.'
• Expose the actual tool registration code from cognis_core.mcp.build_mcp_server() or inline the schema definition so the full parameter and output structure is visible for review.