A native, GPU-rendered control plane for AI coding agents, OpenSquirrel is built with Rust and GPUI for macOS users. It enables running multiple AI coding assistants (Claude Code, Codex, Cursor Agent, OpenCode) side-by-side in a responsive tiled layout. Key benefits include automatic agent delegation, remote machine targeting via SSH, and persistent multi-turn sessions, optimizing productivity for developers leveraging multiple AI tools without the overhead of Electron.

How It Works

OpenSquirrel employs Rust and the GPUI framework for a native, GPU-rendered user interface, avoiding Electron for enhanced performance. Its core features a multi-agent grid that automatically arranges agents based on their count. A coordinator agent can delegate tasks to sub-agents, which return condensed results. The system supports targeting local or remote machines via SSH and integrates with MCP servers. Sessions are persistent, saving agent state, transcripts, and scroll positions, allowing interrupted turns to be resumed. Output from supported runtimes is parsed into structured markdown, including code blocks and diffs.

Quick Start & Requirements

• Primary install / run command: Build from source using cargo build --release and run the executable.
• Non-default prerequisites and dependencies: Rust toolchain, macOS operating system, Metal GPU.
• Notes: Building a .app bundle requires manual steps and may necessitate direct binary execution if agents are not found due to the .app bundle not inheriting the shell's PATH.

Highlighted Details

• Native, GPU-rendered UI built with Rust and GPUI, eliminating Electron.
• Multi-agent grid layout with automatic arrangement for 1, 2, or 4 agents.
• Coordinator/worker delegation model for focused task execution.
• Remote machine targeting capabilities via SSH.
• Persistent sessions that save agent state, transcripts, and scroll positions across restarts.
• Structured markdown rendering for agent output, supporting code blocks and diffs.

Maintenance & Community

The provided documentation does not specify notable contributors, sponsorships, community channels (e.g., Discord, Slack), or a public roadmap.

Licensing & Compatibility

• License type: MIT.
• Compatibility notes: The permissive MIT license generally supports commercial use and linking with closed-source projects.

Limitations & Caveats

The project is currently limited to macOS due to its reliance on Metal GPU rendering via GPUI. A Rust toolchain is required for building. When using the .app bundle, direct binary execution might be necessary for agent discovery if shell PATH is not inherited.