# aichat-dev-docker This is a `sigoden/aichat` configuration to be used with docker. ## Installation 1. Install `docker`, `docker-compose`, `docker-buildx` in your environment 2. Clone this repository and all of its submodules: ```bash git clone --recurse-submodules https://git.kug.is/aichat-dev-docker.git ``` 3. Build `llm-functions-docker`. Refer to the [llm-functions-docker/README.md](llm-functions-docker/README.md) inside the project for detailed guide. 4. `cd` into this project directory: ```bash cd aichat-dev-docker ``` 5. Adjust `config/config.yml` and `config/roles/tools.md` to your preferences or keep the defaults. Aichat is configured to use Deepinfra as model host, and role `tools` for tool usage. 6. Copy `docker-config-example.yml` to `docker-compose.yml` and adjust it to your preferences or keep the defaults. It will use directories `work`, `projects` and `config` relative to the project directory. To integrate it into a host installation, you might want to mount `projects` to your local projects folder, and `config` to `~/.config/aichat` to use the same config on host and docker. 7. Collect required API keys and place them in `.env` files. `config/.env` example: ```conf DEEPINFRA_API_BASE=https://api.deepinfra.com/v1/openai DEEPINFRA_API_KEY=[...] ``` `DEEPINFRA` being the client name as in `config/config.yml`. `llm-functions-docker/.env` example: ```conf TAVILY_API_KEY=[...] ``` Optional. Required, if you want to use `web_search` tool using Tavily. 8. Build the docker image: ```bash docker compose build ``` ## Usage 1. Start the docker container: ```bash docker compose up -d ``` 2. Run aichat inside the docker container. If you want to use tools, you have to enable the role explicitly: ```bash docker exec -it aichat aichat --role tools ``` If you dont want to use tools at all, you can omit this role (or define a different one): ```bash docker exec -it aichat aichat ```