An open-source coding and research agent built around a persistent Python REPL, recursive subagents, and self-refining harness state for long-running tasks.
abcwyc/pi-agent-desktop
A standalone macOS and Windows desktop app that wraps the Pi AI coding agent with a GUI, so you can run it without terminal setup.
- Stars
- 811
- Forks
- 92
- Open issues
- 4
- Last commit
- 24 hours ago
What is it?
Pi Agent is a local AI agent desktop app that packages the agent capabilities of the pi project into an installable application. It lets you browse and resume past Pi sessions grouped by project, and chat with the agent in a desktop window where thinking, tool calls, context usage, cost, and compaction state are visible.
The app can branch a conversation from any earlier message or fork it into an independent session. It also covers model and API key management, custom model configuration, skills and plugins, Git worktree switching, and a file browser with previews for code, diffs, Markdown, images, audio, PDF, and DOCX. Scheduled tasks run saved prompts in new sessions on an hourly, daily, weekly, cron, or one-time basis while the app is open.
The desktop package bundles the Next.js server, Node.js runtime, and current Pi SDK, so no separate terminal or Node.js install is needed. Updates are delivered as complete signed builds from Settings. Existing Pi data in ~/.pi/agent is read by default.
Key features
- 01Browse and resume past Pi sessions organized by project.
- 02Chat with the agent in real time, with tool calls, context usage, and cost visible.
- 03Branch from earlier messages or fork conversations into independent sessions.
- 04Manage models, API keys, skills, and plugins from the settings panel.
- 05Run scheduled prompts in new sessions while the app is open.
- 06Preview code, diffs, Markdown, images, audio, PDF, and DOCX files.
Use cases
- Using Pi's coding agent on macOS or Windows without setting up a terminal or Node.js.
- Reviewing and resuming earlier agent sessions across multiple projects.
- Running recurring read-only prompts, such as daily code checks, on a schedule.
- Working across Git worktrees within one desktop window.
Topics
FAQ
- What is pi-agent-desktop?
- A standalone macOS and Windows desktop app that wraps the Pi AI coding agent with a GUI, so you can run it without terminal setup. Pi Agent is a local AI agent desktop app that packages the agent capabilities of the pi project into an installable application. It lets you browse and resume past Pi sessions grouped by project, and chat with the agent in a desktop window where thinking, tool calls, context usage, cost, and compaction state are visible.
- Is pi-agent-desktop free to use? What license does it have?
- pi-agent-desktop is open source under the MIT license. Check the license terms before using it in commercial products.
- Is pi-agent-desktop still maintained?
- Actively maintained. The last commit to pi-agent-desktop was on Oct 17, 2026.
- What can I use pi-agent-desktop for?
- Using Pi's coding agent on macOS or Windows without setting up a terminal or Node.js. Reviewing and resuming earlier agent sessions across multiple projects. Running recurring read-only prompts, such as daily code checks, on a schedule. Working across Git worktrees within one desktop window.
- Where can I get pi-agent-desktop?
- The source code and installation instructions are on GitHub: https://github.com/abcwyc/pi-agent-desktop
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
An open-source AI coworker for your desktop that completes finished tasks like security reviews, documents, and inbox triage, with approval gates and a sandbox.
Open-source local AI voice studio for cloning voices, generating speech, dictating into any app, and giving AI agents a voice.
Incremental data pipeline framework in Python with a Rust core that keeps vector indexes and knowledge graphs for AI agents continuously in sync with changing sources.
A self-hosted personal AI assistant that runs on your own devices and answers you through chat apps like WhatsApp, Telegram, Slack, and Discord.
A software development methodology and composable skills library that guides coding agents through brainstorming, planning, TDD, and subagent-driven implementation.