AI-automated-Aspen

# AI-automated-Aspen This project aims to integrate Aspen series software with LLMs and MCP tools ## Requirements - Local docker-based MCP tools (refer to https://github.com/modelcontextprotocol/servers/tree/main). See docs in Github and Filesystem servers to set up docker-based MCPs - GitHub API key: see https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/managing-your-personal-access-tokens - Python 3.11+ (this is for later integration with Langgraph Studio) ## Quick start 1. Python environment setup: `pip install -r requirements.txt` 2. Configuration: `configs/mcp_config.json` contains the configuration for the MCP tools. 1. Add your GitHub API key to the `env` section. ```json "env": { "GITHUB_PERSONAL_ACCESS_TOKEN": "<your_github_api_key>" }, ``` 2. Add the path to the `data` directory in the `filesystem` section. Please note that the path should be the absolute path to the local data directory. e.g., "C:\\Users\\...\\data". This folder will be the folder that the agents read from and write to. ```json "filesystem": { "command": "docker", "args": [ "run", "-i", "--rm", "--mount", "type=bind,src=<your_absolute_path_to_local_data_directory> ,dst=/projects/data", "mcp/filesystem", "/projects" ], "transport": "stdio" } ``` ## Usage: Training data collection The very first step of this project is to collect as many `.BKP` (Aspen Backup files) as possible and as diversed as possible. The current strategy is to use MCP tools, including GitHub and Serp tool, to locate `.BKP`files on the internet and provide according documentations on the `.BKP` files. It takes the following steps: 1. run `python -m src.data_collection_github_agent.graph` to create lists of `.BKP` files from GitHub. The agent will automatically create a list of `.BKP` files and save it to the `data` directory. 2. run `python -m scripts.create_download_list` to create a list of download links for the `.BKP` files. 3. use third party tool to download the `.BKP` files from the download links. Recommend using `aria2` to download the files if you are on linux systems and `Xunlei` on windows systems.