Isolated worktrees for each session
AO creates separate workspaces for worker sessions so files, branches, and pull requests stay isolated while the agent does the coding.
Agent Orchestrator
Agent Orchestrator is a desktop IDE for supervising AI coding agents, organizing workspaces, pull requests, CI, and reviews in one place. Free and open source under Apache 2.0.
What Agent Orchestrator does
Agent Orchestrator (AO) is a desktop IDE for supervising AI coding agents. It organizes worker sessions into isolated workspaces, keeps their conversations and terminal sessions accessible, and connects the work they produce to GitHub pull requests, reviews, CI checks, and merge status.
The product is designed to help teams and individual builders run several coding agents at once without losing track of branches or follow-up work. AO acts as the control layer around the agents: it shows project and session state, routes review feedback back to the right worker, and supports a workflow from task creation to merge and cleanup.
Core features
Central session board
A board view shows work across states such as pending work, in review, ready to merge, and merged, with branch, PR, and agent details on the card.
Feedback loop for CI and reviews
The system tracks review comments and CI failures on PRs it opens and routes the issues back to the session that owns the branch.
Chat and terminal session modes
Sessions can run in a structured Chat UI or the agent harness's native terminal experience, with support for switching between them on compatible Claude Code and Codex sessions.
Agent supervision and handoff controls
AO can launch reviewer agents, send follow-up instructions to the correct worker, and keep activity, approvals, usage, and conversation history in one place.
Mobile monitoring and notifications
The app includes a desktop companion experience and an opt-in mobile listener for monitoring runs and receiving notifications from your home network.
Common ways to use AO
Run multiple agents without mixing branches
Use AO when one project needs several coding agents working in parallel on separate tasks, with each session kept in its own worktree and status board entry.
Track PR progress and review state
Use AO to follow a pull request from first commit to merge, while keeping CI status, review comments, and readiness to merge visible in the app.
Close the loop on CI failures and review comments
Use AO when review feedback or broken checks should be routed back to the worker that owns the branch instead of being handled manually in a separate tool.
Work in Chat or terminal mode
Use AO when you want to choose between a structured Chat UI and a native terminal session depending on the agent harness and your preferred workflow.
Monitor agents remotely
Use the mobile companion to check session status, open a terminal, and receive notifications when an agent needs attention while you are away from your desk.
Pros and Cons
Pros
- Keeps agent work separated into isolated worktrees, which reduces branch and file collisions.
- Shows session, branch, PR, review, and CI status in one place, making multi-agent work easier to supervise.
- Creates a feedback loop so failed checks and review comments return to the worker that made the change.
- Supports a wide range of agent harnesses, including Claude Code, Codex, Cursor, Aider, and OpenCode.
- Offers both desktop monitoring and an optional mobile companion for follow-up while away from the workstation.
Cons
- The installation path is desktop-first; the docs say the desktop app is the canonical install method.
- GitHub is the only shipped source-control observation path mentioned in the docs, so broader SCM support is not confirmed in the current rewrite.
- The product depends on already-installed agent harnesses and sign-ins, so it is not a fully bundled all-in-one coding agent runtime.
FAQ
Is Agent Orchestrator free to use?
Yes. The site says Agent Orchestrator is free and open source under Apache 2.0.
What kind of work is AO best suited for?
AO works best when each worker has a clear, reviewable outcome such as a bug fix, feature, refactor, test update, documentation change, or review follow-up.
Does AO replace my coding agent or sit on top of it?
The docs describe AO as a desktop IDE for supervising AI coding agents, with support for isolated workspaces, pull requests, CI, reviews, merge conflicts, and notifications. It also offers both Chat and terminal-based session modes, depending on the harness.
Which agents can I use with AO?
The installation docs say the desktop app is the canonical install path and that AO uses the agent binaries and sign-ins already installed on your machine. Supported harnesses include Claude Code, Codex, Cursor, Aider, OpenCode, Droid, Kimi Code, and GitHub Copilot, among others.
Do I need a separate CLI or bundled agent tools to run it?
No. The docs say the desktop app starts and supervises a local Go daemon, the tool does not bundle provider CLIs, and the legacy npm package is frozen at 0.10.0 and no longer recommended for new users.
Quick Facts
- Category
- Developer Tool
- Product type
- Desktop IDE for supervising AI coding agents
- Primary platform
- Desktop app for macOS, Windows, and Linux
- Source domain
- aoagents.dev
- License
- Apache 2.0
- Install path
- Desktop app; legacy npm CLI frozen at 0.10.0
Agent Orchestrator Alternatives
SonOf
SonOf connects to your repo and PM tool, audits your codebase, and turns approved work into shipped tickets with senior engineering review.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repos and verifies changes with compilers, debuggers, sanitizers, and tests.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads in Firecracker micro-VMs with private networking and SDK, CLI, or MCP control.
Biji
Biji is a versatile platform designed to enhance productivity through innovative tools and features.
hob
hob is an independent workspace for coding agents, with local control over sessions, terminals, history, routing, and follow-up work.
Manta AI
Manta AI is an autonomous web app testing tool that maps app behavior, catches regressions, and generates tests from a URL, no scripts or selectors needed.