c4pt0r/pie

Pie is a terminal-based AI coding assistant that you run inside a software project. Once started, you type requests in plain English and the tool reads files, makes edits, runs shell commands, and keeps memory of your preferences across sessions. It is a rewrite of an earlier project called pi, rebuilt in the Rust programming language for performance and reliability. To use it, you point pie at your project folder, connect it to an AI provider by setting an API key, and then have a conversation with it. You can ask things like "summarize this repository", "fix the failing tests", or "add a README example and run the checks". The tool handles the file reading, editing, and command running on your behalf. When a session ends, pie saves the transcript so you can pick up where you left off with the resume option. Pie works with several popular AI services, including Anthropic, OpenAI, Groq, Mistral, Gemini, and OpenRouter. You can also point it at a locally running AI model if you set up a compatible local server. You switch between providers and models at any time from inside the session. One notable feature is triggers. You describe an automation in plain chat, such as "when this file appears, run the tests", and pie converts that into an ongoing rule that checks conditions in the background and fires when they match. This is useful for watching for build artifacts, file changes, or events from external tools. Triggers run in a separate background process so they do not interrupt your main conversation. All session history, memory notes, stored API keys, and configuration live in a folder called .pie in your home directory by default. The project is open source under the MIT license and is built and tested using standard Rust tooling.