apm-mcp-server

# apm-mcp-server: Agile Practice Map MCP Server ## Overview This is the MCP server for interacting with the data of the Agile Practice Map. This server searches for information from the knowledge base of the Agile Practice Map (※) through a large language model. ※ https://www.agile-studio.jp/agile-practice-map ## Tool 1. **query_apm** - Answers questions related to the Agile Practice Map. - Input: - `question` (string): The user's question regarding Agile practices. - Return value: An answer based on the knowledge base of the Agile Practice Map. 2. **list_apm_practices** - Retrieves a list of practice names included in the Agile Practice Map. - Input: None - Return value: A list of practice names. ## Resource 1. **resource://{practice_name}/url** - Returns the URL of the specified practice name. - Input: - `practice_name` (string): The name of the practice (e.g., "daily-scrum"). - Return value: Information about the practice name and URL `{ "practice_name": ..., "url": ... }`. ## Installing the MCP Server ### Installing uv If not already installed, you need to install the `uv` package manager. **Windows:** ``` powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex" ``` **macOS:** ``` curl -LsSf https://astral.sh/uv/install.sh | sh ``` **Linux:** ``` curl -LsSf https://astral.sh/uv/install.sh | sh ``` ### Cloning the Repository ``` # Clone the repository. git clone https://github.com/HappymanOkajima/apm-mcp-server.git ``` ### Setting Up OpenAI API Key Please create a .env file under the apm-mcp-server directory. Example of the `.env` file: ``` OPENAI_API_KEY=sk-xxxxxxxxxxxxxxxxxxxx ``` ## Configuring the MCP Client ### Usage with Claude Desktop Please add the following to `claude_desktop_config.json`. Specify the file paths as absolute paths, including the command. This is a sample based on Windows. ```json { "mcpServers": { "apm-mcp-server": { "disabled": false, "timeout": 60, "command": "c:\\YOUR_UV_PATH\\uv", "args": [ "--directory", "C:\\YOUR_SERVER_INSTALL_PATH\\apm-mcp-server", "run", "-m", "apm_mcp_server" ], "transportType": "stdio" } } } ``` If you are running the MCP server using a Docker container, after building the Docker image, configure the client as follows. The API key should be specified in `claude_desktop_config.json`. Docker build: ``` docker build -t mcp/apm . ``` ```json { "mcpServers": { "apm": { "command": "docker", "args": [ "run", "--rm", "-i", "-e", "OPENAI_API_KEY=sk-xxxxxxxxxxxxxxxxxxxx", "mcp/apm" ] } } } ``` ### Usage with Cline Please consult with Cline chat for installation and configuration. ## Project Structure - `apm_mcp_server/`: Main server code - `main.py`: Entry point for the MCP server - `rag_chroma/`: RAG implementation using Chroma DB - `data/`: Contains source data and vector database - `apm.txt`: Source text data for the Agile Practice Map - `chroma_db/`: Data for the Agile Practice Map - `tools/`: Utility scripts - `populate_db.py`: Script to create a vector database from source data ## Database To simplify configuration, a ChromaDB database file containing APM information is stored directly under the data directory. ## Database Construction Tool A tool for constructing a vector database from content URLs or text files is included. ``` uv run -m tools.populate_db ``` This will create a vector database in `data/chroma_db/`. Therefore, by using this tool, you can create and add knowledge not only from APM but also from web pages and text files. ## License This MCP server is released under the MIT License. This allows you to freely use, modify, and distribute the software in accordance with the terms of the MIT License. For more details, please refer to the LICENSE file in the project repository.