packkit-mcp is the Model Context Protocol (MCP) server for Packkit, used to let AI agents scaffold, upgrade, and compose modern projects as native tools.
Purpose The server is aimed at AI clients that support MCP (such as Claude Desktop, Cursor, VS Code, etc.), enabling these agents to directly call Packkit's various generators to create or upgrade projects without manually executing commands or writing scripts.
Main capabilities
- Multilingual project generation: all Packkit generators are uniformly wrapped through the @packkit/core protocol, so the same set of tools can work across languages.
- JavaScript / TypeScript: create-packkit, supporting libraries, CLI, services, SPA, monorepo.
- Python: create-packkit-py, supporting libraries, CLI, worker, HTTP services.
- Go: create-packkit-go, supporting libraries, CLI, worker, HTTP services.
- Full-stack composition: provides the compose_fullstack tool, which can splice a static frontend and backend service from any generator into a single full-stack repository, for example React + FastAPI.
- Extensible architecture: adding a new language later only requires adding one line registry.register(...) in the server, and the tools exposed externally do not need to change; all generation processes are protocol-driven, and the server itself does not contain language-specific logic.
How to use Point your MCP client at the published package (no installation required, npx will fetch it automatically). The typical call flow is: list_generators → list_presets / get_generator_schema → generate_project.
Use cases Suitable for developers who need to quickly create or upgrade multilingual project structures in an AI agent environment, and for scenarios where they want to combine frontend and backend into a single full-stack repository.
Limitations The materials do not provide specific limitation descriptions, charging information, prices, platform support details, or usage effect data,Not yet provided.
