How to connect Pi Coding Agent to another AI API
The Reddit post shares a setup for connecting Pi Coding Agent to an outside AI API such as Atlas Cloud. It says Pi reads custom provider settings from `~/.pi/agent/models.json`. The example adds an OpenAI-compatible provider with a model ID, API key, token limits, and compatibility options. The post says to run it with `pi --provider atlascloud --model deepseek-ai/deepseek-v4-pro`.
Key points
- The install command shown is `npm install -g --ignore-scripts @earendil-works/pi-coding-agent`.
- The installed command is `pi`, not `pi-coding-agent`.
- Pi reads custom provider settings from `~/.pi/agent/models.json` on macOS/Linux.
- The example registers an Atlas Cloud DeepSeek model as an OpenAI-compatible provider.
- The config includes `baseUrl`, `apiKey`, `contextWindow`, `maxTokens`, and compatibility options.
Quick term guide
- coding agent
- An AI tool that writes or edits code from a person’s instructions.
- custom provider
- A user-added connection to an AI model service that is not built in by default.
- OpenAI-compatible
- It uses an API shape similar to OpenAI’s, so tools can connect with fewer changes.
- token limits
- Token limits are the maximum amount of text an AI model can handle at one time.
- DeepSeek-V4
- An open-source large language model that can be run on a personal computer without paying for cloud API access.
- contextWindow
- The maximum amount of text the AI can read and use at one time.
- token usage
- Token usage is a count of how much text an AI tool processes.
- coding-agent
- An AI tool that can help write, edit, and reason about code across a task.