mcp-redteam is a red-team testing framework, not a production MCP server. Its 7 fixture tools are intentionally designed to demonstrate security vulnerabilities (unannotated destructive tools, rug-pull descriptions, authorization bypasses). Tool definitions are minimal and deliberately flawed for testing purposes. All tools lack parameter descriptions entirely. Most tools have bare-minimum descriptions (10-30 chars). No input schemas are visible in the provided code, only parameter names and types are inferred from fixture definitions. The server is not intended for production use; it's a security testing harness.

• Add explicit JSON Schema input definitions for all tools with type and description for every parameter. Example: {"type": "object", "properties": {"city": {"type": "string", "description": "City name (e.g., 'San Francisco')"}}}
• Expand tool descriptions to 50-200 characters, explaining WHAT the tool does, WHEN to use it, and what it returns. Current descriptions are too terse for LLM selection.
• Document output schemas for all tools. Specify returned fields, types, and whether results are paginated. Include example responses.
• Add parameter descriptions explaining format, constraints, and valid values. For 'filter' in delete_records, specify the filter syntax and examples.
• For destructive tools (delete_records, purge_cache), add error handling guidance: 'Returns 400 if filter is invalid. Returns 409 if records are locked. Retry with exponential backoff on 503.'