Decompose MCP

Well-documented deterministic text classifier with clear tool descriptions and transparent security model

• +Provides highly specific, actionable tool descriptions with clear parameters and use cases
• +Excellent security documentation with explicit SSRF protections and source traceability
• +Strong performance characteristics and deterministic behavior reduce LLM context window needs
• -Over-scoped permissions: filesystem read/write and shell execution not justified by MCP tools
• -No examples showing actual agent interaction patterns or multi-step workflows