Our take
Newless than six months old and changing fast- Activity
- 84
- Community
- 34
- Issues
- 53
- Pull requests
- 84
- Merges more pull requests than 96% of the projects we track
- Releases more often than 85% of the projects we track
diri is a fast-moving, early-stage desktop coordinator with substantial recent development activity, but its development is heavily concentrated in one contributor.
Good fit if
- You already use terminal coding agents and want a shared view of their status and changes.
- You work on macOS, or are willing to evaluate the more limited Linux beta.
- You want agent tasks separated by Git worktrees and are comfortable with an AGPL-3.0 project.
Look elsewhere if
- You need a mature project with development distributed across contributors, since most contributions currently come from one person.
- You require stable Linux support, as Linux is explicitly beta and packages are not included in every release.
- Your workflow does not involve running multiple coding agents or reviewing their work in parallel.
All health signals
| Last commit | 2026-11-04 (today) |
|---|---|
| Commits, last 90 days | 100+ |
| Releases, last 12 months | 45 (latest v0.11.2, 2026-11-04) |
| Contributors | 18 (top contributor: 92% of commits) |
| Issues closed, last 90 days | 2+ |
| Pull requests merged, last 90 days | 95 (typically merged in 0 days) |
| Project age | 3 months |
Checked on 2026-11-04 with the GitHub API.
Overview
diri is a native desktop app built with Rust and GPUI for managing coding agents in parallel. It addresses the coordination overhead of running many agents by showing their status in one place, separating tasks into Git worktrees, and keeping changes available for review alongside each session.
It runs agent CLIs already installed on your machine and supports macOS, with Linux in beta. Agents can run locally or on an SSH host you control. Claude Code and Codex have deeper status and resume integrations; other supported terminal agents can also be used.
Key Features
- Shows live session status and notifications so you can identify agents waiting for input or finished.
- Provides an MCP server that lets agents spawn other agents, delegate tasks, read output, and answer questions.
- Can give each task its own Git worktree and branch to keep parallel changes separate.
- Brings diffs, staging, commits, and pull request checks into the session review workflow.
- Keeps terminal sessions running independently of the app, including when the app closes or the Engine restarts.
- Can run agents locally or on an SSH host controlled by the user, without a hosted relay.
Use Cases
- A developer coordinating several Claude Code or Codex sessions can monitor which ones need attention without switching between many terminals.
- A small team exploring parallel implementation can assign separate worktrees to agent tasks and review each set of changes in context.
- An experienced agent user can use the MCP server to delegate subtasks from one agent to others and inspect their results.
- A developer who wants to run agent CLIs on a machine they control can use diri as a local or SSH-hosted session manager.
What you need
Detected in the repository
- Rust (main language on GitHub)
- Automated checks on GitHub Actions
License in plain words
AGPL-3.0network copyleft
- Commercial use: yes
- Modify and redistribute: yes
- You must keep: the license notice
- Share your changes: yes, the full source, also when people only use it over a network (e.g. as a hosted service)
- Includes an explicit patent grant from the contributors:
A summary, not legal advice: the LICENSE file is what applies.
Getting Started
On macOS 15 or newer, install with Homebrew:
brew install --cask diri
See the quickstart and README for setup details. Linux users should consult the Linux guide.
Alternatives
- orca: Orca supports mobile and SSH control and more agent CLIs, while diri focuses on a macOS desktop with deeper Claude Code and Codex integrations.
- agent-orchestrator: Agent Orchestrator emphasizes project planning, Kanban tracking, CI, and reviews, while diri focuses on running agents side by side and reviewing changes.
- claude-code-by-agents: Claude Code by Agents coordinates work through threaded rooms across local and remote machines, while diri centers on a desktop workspace for parallel agents.
| Project | Language | License | Stars | Status |
|---|---|---|---|---|
| diri | Rust | AGPL-3.0 | 483 | New |
| orca | TypeScript | MIT | 101k | Active |
| agent-orchestrator | Go | Apache-2.0 | 13.3k | Active |
| claude-code-by-agents | Swift | MIT | 892 | Quiet |
Considerations
The project is new and changing quickly. Recent activity includes at least a hundred commits in the measured period and dozens of releases, while the contributor base is relatively small and most code comes from one person. Recent pull request merging is active, but issue closure data is limited, so users should assess support and maintenance needs before depending on it for critical workflows. Linux support remains beta, and the project is licensed under AGPL-3.0, which teams should review for their intended use.
Found this useful?
Share it with someone who would like diri.