What problem does it solve? Building agent capabilities requires consistent, stateless tools that follow MCP principles, but ad-hoc tool code often lacks structure, typed inputs, and proper registry wiring. This Skill provides the conventions and patterns for creating tools that agents can reliably call. ## Core Features & Use Cases - MCP Tool Generation: Create raw tool functions in shared/mcp/tools/ that accept typed parameters and return structured dicts. - LangChain Wrapper Creation: Generate @tool decorated wrappers in shared/mcp/server.py that expose tools to agents with LLM-friendly descriptions. - Registry & Agent Binding: Add tools to ALL_TOOLS and map them to agent types (research, analysis, generator, code, monitor, chat) in TOOL_REGISTRY, including CrewAI conversion via Tool.from_langchain(). - Use Case: When adding a new capability like a weather API integration, use this Skill to scaffold the raw function, the LangChain wrapper, and the registry entries so research and generator agents can invoke it immediately. ## Quick Start Create a new MCP tool that fetches stock prices and register it for the research and analysis agents.
