Single tool 'memtriage_scan' has a reasonable description (95 chars) but critically lacks a documented input schema in the source code. The FastMCP decorator infers the schema from the Python type hint (target: str), but no explicit JSON Schema with type, description, or constraints is visible. Parameter 'target' has no description in the tool definition. Output schema is undocumented, the tool returns to_json(scan(target)), but the structure of that JSON is not declared. No error handling guidance, no validation rules, no recovery hints. Naming is acceptable (verb_noun pattern: 'memtriage_scan'), but the tool does only one thing and lacks the depth expected of production tools.

• Add explicit JSON Schema to memtriage_scan input: declare 'target' as type string with minLength, pattern (e.g., valid path), and description: 'Path to memory dump file or directory to scan. Must be readable and contain text or binary artifacts.'
• Document output schema in tool description or as a separate schema block. Example: '{"tool": "memtriage", "findings": [{"id": string, "sev": string, "where": string}], "score": number}'
• Add parameter description for 'target': 'Absolute or relative path to a memory dump export file or directory. Supports recursive directory traversal. Returns error if path does not exist or is not readable.'
• Implement error handling with actionable messages: 'Path not found: /path/to/dump. Verify the path exists and is readable.' or 'Permission denied: /path/to/dump. Check file permissions.'
• Migrate from STDIO to HTTP or Streamable HTTP transport to enable remote access and compatibility with hosted MCP clients.