PI-Desktop

Overview

PI-Desktop is a local-first desktop workspace for AI agents. It brings projects, agents, models, plugins, and workflows into one persistent environment, supports Agent, Plan, and Goal work modes, and uses a plugin-powered core to extend the agent, workspace, and platform.

Key Features

• Persistent workspace for projects, sessions, reviews, previews, and agents, with sessions that can continue across app launches.
• Plugin system for commands, panels, floating widgets, work panel views, agent tools, completion, skills, themes, MCP servers, services, and a plugin message bus.
• Agent orchestration through Subagents for independent work and Worker Sessions for longer parallel flows.
• Model-agnostic configuration with cloud models, local models, custom gateways, OpenAI-compatible APIs, Ollama, and LM Studio, with per-model settings and in-session switching.
• Local-first data handling: projects, sessions, settings, and logs stay local by default; credentials use the OS keychain; model requests go directly to the configured provider.
• Permission layer for privileged operations with Allow, Ask, and Deny decisions, so each session can have its own level of autonomy.

Use Cases

• Everyday AI-assisted development where the agent reads code, edits files, runs commands, tests, and iterates.
• Refactors and high-risk changes where Plan mode reviews an implementation plan before execution.
• Complex, long-running work where Goal mode locks the objective and acceptance criteria and Worker Sessions run parallel workstreams.
• Building plugin-powered workspace products such as voice agents, DevOps consoles, GitHub workspaces, data assistants, or multi-agent control centers.

Getting Started

• Download and install PI-Desktop for macOS, Windows, or Linux.
• Connect a model by configuring a provider.
• Open a local project repository.
• Start working in Agent, Plan, or Goal mode.
• Developers can run from source with Node.js >=22.19, pnpm >=10, and a stable Rust toolchain; the documented flow is clone the repository, run pnpm install, build host-core with cargo build -p host-core, run pnpm build:js, then pnpm dev.

Deployment & Requirements

• Packages are available for macOS Apple Silicon and Intel, Windows x64, and Linux x64/ARM64.
• Linux packages require glibc 2.35+, with common support for Ubuntu 22.04+, Debian 12+, and Fedora 36+.
• Running from source requires Node.js >=22.19, pnpm >=10, and a stable Rust toolchain.
• macOS releases are signed with a Developer ID certificate and notarized by Apple.

Before You Adopt

• License: LGPL-3.0. Review its terms before using, modifying, or distributing the project.
• The 0.21.x release line is labeled Early Preview, indicating pre-1.0 maturity.
• When a remote model is used, the context required for the request is sent directly to that provider.