Devtoolwatch BoardPI-Desktop review
PI-Desktop is a local-first desktop workspace for AI coding agents, built on Electron, a Rust host core, the pi agent harness, and plugins.
TL;DR
PI-Desktop is a desktop app for macOS, Windows, and Linux that gives AI agents a persistent workspace of projects and sessions. It is for developers who want an agent outside any specific IDE or terminal, with replaceable models. It stands apart through plugins that extend both the agent and the desktop, and through Subagents and parallel Worker Sessions.
- 7.6kGitHub stars
- +344new stars, week of Oct 25
- #21in Coding agents this week
PI-Desktop at a glance
- Category
- Coding agents
- Maintained by
- Lan (individual, @vastsa on GitHub)
- License
- LGPL 3.0
- Language
- TypeScript, JavaScript, Rust
- Repository
- github.com/vastsa/PI-Desktop
- Website
- pi-docs.aiuo.net
- Latest release
- v0.21.0 on GitHub, Oct 24, 2026
- First published
- Mar 22, 2023
What PI-Desktop does
- PI-Desktop runs coding agents in three modes: Agent for direct execution, Plan for reviewing an approach first, and Goal for outcome-driven long tasks.
- PI-Desktop delegates work to Subagents with their own context, or coordinates parallel Worker Sessions from one parent Session.
- PI-Desktop works with OpenAI, Anthropic, OpenAI-compatible APIs, custom gateways, Ollama, LM Studio, and local models, switchable within a Session.
- PI-Desktop plugins can add commands, panels, floating widgets, work panel views, agent tools, skills, themes, MCP servers, and background services.
- PI-Desktop keeps persistent Sessions that you can pin, archive, branch, search, and continue across app launches.
- PI-Desktop can import local sessions from Claude Code, Codex, OpenCode, and Pi.
- PI-Desktop routes privileged agent operations through a permission layer with Allow, Ask, or Deny decisions per Session.
How it works
PI-Desktop is an Electron desktop app with a Rust host core. Its agent runtime uses pi-ai, pi-agent-core, and pi-coding-agent from the pi ecosystem. Projects, sessions, settings, and logs stay local, and credentials go in the OS keychain. Model requests go directly to your configured provider. Plugins install as .piplug packages or through a marketplace.
How to install and start using PI-Desktop
Check glibc version on Linux
ldd --versionRequirements
- Linux packages require glibc 2.35+, for example Ubuntu 22.04+, Debian 12+, or Fedora 36+.
- Running from source requires Node.js >=22.19, pnpm >=10, and a stable Rust toolchain.
- A configured model provider or a local model is needed to run agents.
First steps
- Download PI-Desktop from the GitHub releases page for macOS, Windows, or Linux.
- Connect a model by configuring a provider, such as OpenAI, Anthropic, Ollama, or a custom gateway.
- Open a project by choosing a local repository.
- Start working in Agent, Plan, or Goal mode.
Who PI-Desktop is for
- Developers who want a coding agent workspace independent of any IDE or terminal.
- Teams splitting complex work across parallel agent sessions, such as frontend, backend, tests, and review.
- Developers who switch between cloud and local models or keep data local.
- Developers building agent plugins, MCP servers, skills, or tools on a host platform.
Limitations to know
- PI-Desktop is in Early Preview, with the 0.21.x release line current and 0.22.0 still in beta.
- PI-Desktop Linux builds need glibc 2.35 or newer, which excludes older distributions.
- PI-Desktop sends request context directly to remote model providers when you use a cloud model.
Is PI-Desktop growing? Usage data
In the week of Oct 25–31, 2026, PI-Desktop ranked #36 on the Devtoolwatch Board and #21 in Coding agents, with a momentum score of 80 out of 100.
Show the weekly numbers
| Week of | New stars |
|---|---|
| Aug 9, 2026 | 3 |
| Aug 16, 2026 | 7 |
| Aug 23, 2026 | 12 |
| Aug 30, 2026 | 13 |
| Sep 6, 2026 | 3k |
| Sep 13, 2026 | 1.4k |
| Sep 20, 2026 | 1.3k |
| Sep 27, 2026 | 483 |
| Oct 4, 2026 | 346 |
| Oct 11, 2026 | 355 |
| Oct 18, 2026 | 294 |
| Oct 25, 2026 | 344 |
The Board ranks tools by four signals each week: new stars, new stars for the repo's size, added downloads and download growth beyond the market. How the ranking works.
PI-Desktop alternatives
Other AI coding agents on the Board in the week of Oct 25–31, 2026, by momentum:
| Rank | Tool | Stars | Downloads/wk |
|---|---|---|---|
| #1 | CindyCindy is an open-source AI agent app for desktop and mobile that combines Claude Code, Codex and Pi across your connected devices. | 8.8k | – |
| #2 | OpenMausBotOpenMausBot is an open-source desktop chat app where each contact is a local Claude, Codex, or Grok CLI agent with its own computer. | 9.6k | – |
| #3 | OrcaOrca is a desktop agent development environment for running Claude Code, Codex, and other CLI coding agents in parallel, each in its own… | 111.7k | – |
| #4 | OpenBotOpenBot is an open-source, self-hosted template for AI coworkers that each get their own browser, files and granted tools, with every… | 9.1k | – |
| #5 | Agent ReachAgent Reach is a Python CLI that lets AI agents search and read platforms such as Twitter, Reddit, YouTube, GitHub, Bilibili and… | 107.2k | – |
| #6 | Skill RecorderSkill Recorder is a desktop app that records your on-screen work session and turns it into a reusable agent Skill or Automation. | 5.6k | – |
Recent PI-Desktop releases
- v0.22.0-beta.3 (pre-release), Nov 1, 2026
- v0.22.0-beta.2 (pre-release), Oct 28, 2026
- v0.22.0-beta.1 (pre-release), Oct 26, 2026
- v0.21.0, Oct 24, 2026
- v0.21.0-beta.8 (pre-release), Oct 23, 2026
PI-Desktop FAQ
Is PI-Desktop free and open source?
Yes. PI-Desktop's source code is on GitHub (vastsa/PI-Desktop) under the LGPL 3.0 license, so you can use it for free.
How do I install PI-Desktop?
Run ldd --version. The install section above has the other options from the project's README.
Is PI-Desktop still maintained?
Yes. The vastsa/PI-Desktop repository was last updated on Nov 1, 2026, and the latest release, v0.21.0 on GitHub, came out on Oct 24, 2026.
How popular is PI-Desktop?
PI-Desktop has 7.6k GitHub stars as of Nov 1, 2026. It gained 344 new stars in the week of Oct 25–31, 2026, against a 12-week average of 626 a week. That made it #21 among AI coding agents on the Devtoolwatch Board.
What are the best alternatives to PI-Desktop?
Other AI coding agents on the Devtoolwatch Board in the week of Oct 25–31, 2026, by momentum: Cindy (#1), OpenMausBot (#2), Orca (#3), OpenBot (#4), Agent Reach (#5). The alternatives table above has their numbers.
Does PI-Desktop require an account or send telemetry?
PI-Desktop does not require an account and has no PI-Desktop telemetry. There is no mandatory relay. Projects, sessions, settings, and logs stay local, and API credentials are stored in the OS keychain.
Which models does PI-Desktop support?
PI-Desktop supports OpenAI, Anthropic, OpenAI-compatible APIs, custom gateways, Ollama, LM Studio, and local models. Each model is configured independently, and a Session can switch models at any time.
Can PI-Desktop run multiple agents at once?
PI-Desktop can run multiple agents. Subagents handle independent tasks with their own context and report back. The Session Orchestrator coordinates full Worker Sessions in parallel, and each Worker stays inspectable with a full transcript.
Can I bring sessions from other coding agents into PI-Desktop?
PI-Desktop can import local sessions from Claude Code, Codex, OpenCode, and Pi.
How does PI-Desktop control what the agent is allowed to do?
PI-Desktop sends privileged tool requests through a permission layer that allows, asks, or denies each one. You decide how much autonomy each Session gets.
What is PI-Desktop built on?
PI-Desktop is built on the pi ecosystem, using pi-ai, pi-agent-core, and pi-coding-agent as its agent runtime. It combines Electron with a Rust host core.
Sources
PI-Desktop README badge
Maintainers can show the tool's weekly Board rank in their README:
[](https://index.devtoolwatch.com/tools/vastsa-pi-desktop/)How this page is made: Devtoolwatch writes each review from the project's README, releases, package metadata and website, with an AI model, then checks every claim against the source it cites and drops what the sources don't support. The numbers come from Devtoolwatch's daily GitHub snapshots, GitHub's stargazer history, and the npm and PyPI download APIs. Spotted an error? Email [email protected].