t4ea-mcp

# t4ea-mcp MCP server that provides **Teamcenter Gateway for Enterprise Applications (T4EA)** API documentation to AI assistants. It enables AI coding tools like Claude, Cursor, and Windsurf to look up ITK functions, validate Tcl scripts, and navigate T4EA documentation accurately. ## Why AI assistants frequently hallucinate function names and parameters when writing Teamcenter/T4EA code. This MCP server gives them access to the actual API reference, a function allowlist, and a script linter — drastically reducing errors in generated Tcl/ITK code. ## Installation ### From PyPI ```bash pip install t4ea-mcp ``` ### From source ```bash git clone https://github.com/aldokruger/teamcenter-gateway-mcp.git cd teamcenter-gateway-mcp uv sync ``` **Requirements:** Python 3.10+ ## Quick Start ### Run locally (stdio) ```bash # Installed from PyPI t4ea-mcp # From source uv run python -m t4ea_mcp ``` ### Run as HTTP server ```bash t4ea-mcp --http --port 8000 ``` ### Use the hosted version A public instance is available at: ``` https://t4ea-mcp.fly.dev/mcp ``` ## Connecting to Your AI Client ### Claude Desktop Add to your `claude_desktop_config.json`: ```json { "mcpServers": { "t4ea": { "command": "t4ea-mcp" } } } ``` Or use the remote server: ```json { "mcpServers": { "t4ea": { "url": "https://t4ea-mcp.fly.dev/mcp" } } } ``` ### Claude Code ```bash claude mcp add t4ea -- t4ea-mcp ``` Or remote: ```bash claude mcp add t4ea --transport http https://t4ea-mcp.fly.dev/mcp ``` ### Cursor / Windsurf / VS Code Add to your MCP settings (`.cursor/mcp.json`, etc.): ```json { "mcpServers": { "t4ea": { "command": "t4ea-mcp" } } } ``` ## Tools The server provides 9 tools: | Tool | Purpose | |------|---------| | `resolve_topic` | **Start here** — routes natural language queries to the right docs | | `search_docs` | Grep for exact strings across all documentation | | `get_doc` | Read a specific documentation …