Agent Teams AI｜AI智能体团队编排桌面应用

access：来源确认 · No prerequisites - the app can detect installed Claude Code, Codex, and OpenCode runtimes. You can also connect Cursor, SuperGrok, GitHub Copilot, Z.AI, MiniMax, and Kiro from the UI.

project_type：来源确认 · An orchestration layer for AI agent teams across Codex/Claude/OpenCode/Cursor/Grok/GitHub/Kiro/Z.AI/Xiaomi/MiniMax/Kimi...(300+ models, 200+ LLM providers, free models no auth)

### [Boss, watch the video](https://agentteams.live/)

<p align="center">
  <a href="https://agentteams.live/">
    <img alt="Agent Teams AI - Get a lot done by doing very little" width="100%" src="docs/assets/agent-teams-overview.webp" />
  </a>
</p>

<p align="center">
  <a href="https://github.com/777genius/agent-teams-ai/releases/latest"><img src="https://img.shields.io/github/v/release/777genius/agent-teams-ai?style=flat-square&label=version&color=007ec6" alt="Latest Release" /></a>&nbsp;
  <a href="https://discord.gg/qtqSZSyuEc"><img src="https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fdiscord.com%2Fapi%2Fv10%2Finvites%2FqtqSZSyuEc%3Fwith_counts%3Dtrue&query=%24.approximate_member_count&label=Discord&logo=discord&logoColor=white&color=5865F2&style=flat-square&suffix=%20members" alt="Discord" /></a>&nbsp;
  <a href="https://trendshift.io/repositories/24276?utm_source=repository-badge&amp;utm_medium=badge&amp;utm_campaign=badge-repository-24276" target="_blank" rel="noopener noreferrer"><img src="https://trendshift.io/api/badge/repositories/24276" alt="777genius%2Fagent-teams-ai | Trendshift" width="125" height="26"/></a>
  <a href="https://agentteams.live/"><img src="https://img.shields.io/badge/site-agentteams.live-8B5CF6?style=flat-square&logo=googlechrome&logoColor=white" alt="AgentTeams" /></a>&nbsp;
  <a href="https://github.com/777genius/agent-teams-ai/actions/workflows/ci.yml?query=branch%3Amain"><img src="https://github.com/777genius/agent-teams-ai/actions/workflows/ci.yml/badge.svg?branch=main&amp;event=push" alt="CI Status (main)" /></a>&nbsp;
</p>

<p align="center">
  <sub>Free cross-platform desktop app for AI agent teams. Start with a free model with no auth - no signup, API key, or card - or connect Claude Code, Codex, OpenCode, Cursor, SuperGrok, GitHub Copilot, Z.AI, MiniMax, Kiro, and many more. For coding and broader project work.</sub>

  <sub>Check also:
  - [universal-agent-plugins](https://universal-agent-plugins.com) - Multi-agent CLI to install, update, repair, and remove Agent Plugins 1.0 and Agent Skills across Codex/ChatGPT, Claude, Cursor, Gemini, Copilot, Qoder, Pi, OpenCode, and more. 1,500+ plugins. SDK for universal plugins (hooks, detect, install/manage) 
  - [review-router-ai](https://github.com/777genius/review-router-ai) - Free, open source CodeRabbit. Powerfull AI code review that runs inside your CI/CD (privacy) using your Claude/Codex/other subscription accounts (pool)/API. Free/self-hosted deployment. Ready github integration. Built using OpenCodeReview. Claude/Codex/OpenRouter/Grok/Kimi/Glm/MiniMax/MiMo...
  - [agent-notifications](https://agent-notifications.com) - 🔔 Cross-platform smart notifications for Claude/Codex/OpenCode/Cursor/Gemini/Qoder/Pi. Desktop alerts, sounds, click-to-focus, and webhooks. macOS, Linux, and Windows.
  </sub>
</p>

<!--
<p align="center">
  <a href="docs/screenshots/overview.jpg">
    <img src="docs/screenshots/overview.jpg" alt="Agent Teams AI workspace overview" width="100%" />
  </a>
</p>
-->

### Screenshots ([watch on our site](https://agentteams.live/#screenshots)):

<table>
<tr>
<td width="50%">
  <a href="docs/screenshots/task-detail-animated.gif">
    <img src="docs/screenshots/previews/task-detail-animated.gif" alt="Agent Teams AI product demo" width="100%" loading="lazy" />
  </a>
</td>
<td width="50%">
  <a href="docs/screenshots/15.png">
    <img src="docs/screenshots/previews/15.webp" alt="Select AI providers and models" width="100%" loading="lazy" />
  </a>
</td>
</tr>
<tr>
<td width="50%">
  <a href="docs/screenshots/1.png">
    <img src="docs/screenshots/previews/1.webp" alt="Kanban board with team messages and agent tasks" width="100%" loading="lazy" />
  </a>
</td>
<td width="50%">
  <a href="docs/screenshots/8.png">
    <img src="docs/screenshots/previews/8.webp" alt="Task details, attachments, and execution logs" width="100%" loading="lazy" />
  </a>
</td>
</tr>
<tr>
<td width="50%">
  <a href="docs/screenshots/12.png">
    <img src="docs/screenshots/previews/12.webp" alt="Full-screen team chat with replies, attachments, mentions, and drafts" width="100%" loading="lazy" />
  </a>
</td>
<td width="50%">
  <a href="docs/screenshots/9.png">
    <img src="docs/screenshots/previews/9.webp" alt="Agent execution log with tool calls" width="100%" loading="lazy" />
  </a>
</td>
</tr>
<tr>
<td width="50%">
  <a href="docs/screenshots/7.png">
    <img src="docs/screenshots/previews/7.webp" alt="Code review with file-level and hunk-level controls" width="100%" loading="lazy" />
  </a>
</td>
<td width="50%">
  <a href="docs/screenshots/4.png">
    <img src="docs/screenshots/previews/4.webp" alt="Team members with roles, runtimes, and models" width="100%" loading="lazy" />
  </a>
</td>
</tr>
<tr>
<td width="50%">
  <a href="docs/screenshots/13.png">
    <img src="docs/screenshots/previews/13.webp" alt="Connected AI coding providers and subscription limits" width="100%" loading="lazy" />
  </a>
</td>
<td width="50%">
  <a href="docs/screenshots/6.png">
    <img src="docs/screenshots/previews/6.webp" alt="Team notification settings and built-in triggers" width="100%" loading="lazy" />
  </a>
</td>
</tr>
<tr>
<td width="50%">
  <a href="docs/screenshots/2.png">
    <img src="docs/screenshots/previews/2.webp" alt="Organization hierarchy with teams, agents, and active tasks" width="100%" loading="lazy" />
  </a>
</td>
<td width="50%">
  <a href="docs/screenshots/14.png?v=1cff3f1128ca">
    <img src="docs/screenshots/previews/14.webp?v=1cff3f1128ca" alt="Token usage, costs, runs, and budget analytics" width="100%" loading="lazy" />
  </a>
</td>
</tr>
</table>

<!-- <p align="center"><sub>Click any preview to open the full-size image.</sub></p> -->

<!--
<img width="1304" height="820" alt="image" src="https://github.com/user-attachments/assets/dea53a01-68b3-4c36-bcf6-e4d1ad4cdb31" />
<img width="2560" height="1604" alt="telegram-cloud-photo-size-2-5213160716557294036-w" src="https://github.com/user-attachments/assets/cbe0e1e7-f73c-4dd7-94e0-c694fcaf491e" />
-->

<!--
<a href="https://agentteams.live/">Watch demo on the site or here:</a>

[demo_new_15s.webm](https://github.com/user-attachments/assets/d78cf5a4-80fe-4a8b-a1db-fb272e18029c)

<details>
<summary>Older demos</summary>

<br />

<table>
<tr>
<td width="50%">

https://github.com/user-attachments/assets/9cae73cd-7f42-46e5-a8fb-ad6d41737ff8

</td>
<td width="50%">

https://github.com/user-attachments/assets/35e27989-726d-4059-8662-bae610e46b42

</td>
</tr>
</table>

</details>
-->

<br />

## Installation

No prerequisites - the app can detect installed Claude Code, Codex, and OpenCode runtimes. You can also connect Cursor, SuperGrok, GitHub Copilot, Z.AI, MiniMax, and Kiro from the UI.

<table align="center">
<tr>
<td align="center">
  <a href="https://github.com/777genius/agent-teams-ai/releases/latest/download/Agent.Teams.AI-arm64.dmg">
    <img src="https://img.shields.io/badge/macOS_Apple_Silicon-.dmg-000000?style=for-the-badge&logo=apple&logoColor=white" alt="macOS Apple Silicon" />
  </a>
  <br />
  <a href="https://github.com/777genius/agent-teams-ai/releases/latest/download/Agent.Teams.AI-x64.dmg">
    <img src="https://img.shields.io/badge/macOS_Intel-.dmg-434343?style=for-the-badge&logo=apple&logoColor=white" alt="macOS Intel" />
  </a>
</td>
<td align="center">
  <a href="https://github.com/777genius/agent-teams-ai/releases/latest/download/Agent.Teams.AI.Setup.exe">
    <img src="https://img.shields.io/badge/Windows-Download_.exe-0078D4?style=for-the-badge&logo=windows&logoColor=white" alt="Windows" />
  </a>
  <br />
  <sub>May trigger SmartScreen - click "More info" -> "Run anyway"</sub>
  <br />
  <sub>Run normally. Administrator mode may be needed only if the app reports a specific OpenCode symlink or permission error.</sub>
</td>
<td align="center">
  <a href="https://github.com/777genius/agent-teams-ai/releases/latest/download/Agent.Teams.AI.AppImage">
    <img src="https://img.shields.io/badge/Linux-Download_.AppImage-FCC624?style=for-the-badge&logo=linux&logoColor=black" alt="Linux AppImage" />
  </a>
  <br />
  <a href="https://github.com/777genius/agent-teams-ai/releases/latest/download/agent-teams-ai-amd64.deb">
    <img src="https://img.shields.io/badge/.deb-E95420?style=flat-square&logo=ubuntu&logoColor=white" alt=".deb" />
  </a>&nbsp;
  <a href="https://github.com/777genius/agent-teams-ai/releases/latest/download/agent-teams-ai-x86_64.rpm">
    <img src="https://img.shields.io/badge/.rpm-294172?style=flat-square&logo=redhat&logoColor=white" alt=".rpm" />
  </a>&nbsp;
  <a href="https://github.com/777genius/agent-teams-ai/releases/latest/download/agent-teams-ai.pacman">
    <img src="https://img.shields.io/badge/.pacman-1793D1?style=flat-square&logo=archlinux&logoColor=white" alt=".pacman" />
  </a>
</td>
</tr>
</table>

## Table of contents

- [Installation](#installation)
- [Table of contents](#table-of-contents)
- [What is this](#what-is-this)
- [Comparison](#comparison)
- [Quick start](#quick-start)
- [FAQ](#faq)
- [Roadmap (new)](#roadmap-new)
- [Vision](#vision)
- [Development](#development)
- [Tech stack](#tech-stack)
  - [Terminal Platform integration](#terminal-platform-integration)
  - [Debug teammate runtimes](#debug-teammate-runtimes)
  - [Build for distribution](#build-for-distribution)
  - [Scripts](#scripts)
- [Contributing](#contributing)
- [Partnerships](#partnerships)
- [Security](#security)
- [License](#license)

## ❤️ Sponsors

<p>
  <a href="https://fluxionai.world/register?source=github&campaign=github-agent-teams&promo=AGENTTEAMS"><img src="docs/assets/sponsor-fluxion.png" alt="Sponsor: Fluxion AI - one gateway to the world's leading AI models. $3 signup credit" /></a>

  
  <a href="#partnerships"><img src="docs/assets/sponsor-become.png" alt="Become a sponsor - partner with or sponsor Agent Teams" width="26%" /></a>
</p>

## What is this

An orchestration layer for AI agent teams across Codex/Claude/OpenCode/Cursor/Grok/GitHub/Kiro/Z.AI/Xiaomi/MiniMax/Kimi...(300+ models, 200+ LLM providers, free models no auth)
- **Assemble your team** — create agent teams (using 1 prompt) with different roles that work autonomously in parallel
- **Sit back and watch** — tasks change status on the kanban board while agents handle everything on their own
- **Agents talk to each other** — they communicate, create and manage their own tasks, review, leave comments
- **Cross-team communication** — agents can fully communicate across different teams; you can configure or prompt them to collaborate and message each other between teams
- **Built-in review workflow** — easily see how agents review each other's tasks to make sure everything went exactly as planned
- **Token analytics and budgets** — see input, output, cache, and reasoning usage across teams, agents, tasks, projects, models, runtimes, sessions, commands, and runs. Spot expensive work, track trends and forecasts, set monthly token or estimated-cost budgets, and get alerts at 80% and 100%
- **Organizations and global overview** — group teams into departments, squads, or any nested structure. Track every organization on one live map with team and agent status, task progress, dependencies, delegation, and cross-team communication
- **Stay in control** — send a direct message to any agent (DM or group chat), drop a comment on a task, or pick a quick action right on the kanban card whenever you want to clarify something or add new work
- **Flexible autonomy** — let agents run fully autonomous, or review and approve supported tool actions one by one (you'll get a notification) — configure the level of control that fits your security needs
- **Task-specific logs and messages** — clearly see agent/runtime logs (tools), actions and messages in isolation for each individual task, making it easy to trace what happened for any assignment
- **Integrated terminal workspace** — run commands in a visual PTY scoped to each team and project, with local shell tabs, persistent history, autocomplete, and terminal settings
- **Solo mode** — one-member team: a single agent that creates its own tasks and shows live progress. Saves tokens; can expand to a full team anytime
- **Multi-provider orchestration** — start with a free model with no auth, auto-detect available Claude Code, Codex, and OpenCode runtimes, or connect Cursor, SuperGrok, GitHub Copilot, Z.AI, MiniMax, Kiro and more, using the subscriptions or API keys you already have
- **Review changes like in Cursor** — see what code each task changed, then approve, reject, or comment

<details>
<summary><strong>More features</strong></summary>

- **Live process section** — see registered background services, stop them, and open their URLs directly in the browser

- **Per-agent CPU/RAM history** — see how many local resources each live teammate uses over time

- **Task creation with attachments** — send a message to the team lead with any attached images. The lead will automatically create a fully described task and attach your files directly to the task for complete context.

- **Auto-resume after rate limits** — when the lead hits a Claude rate limit and the reset time is known, the app can automatically nudge the lead to continue once the cooldown has passed

- **Deep session analysis** — detailed breakdown of each agent session: agent output, tool calls, bash commands, and subprocesses

- **Smart task-to-log/changes matching** — automatically links session logs/changes to specific tasks

- **Advanced context monitoring system** — see what consumes context: user messages, project instructions, tool outputs, available thinking summaries, and team coordination. View token usage, context-window share, and estimated cost by category.

- **Recent tasks across projects** — browse the latest completed tasks from all your projects in one place

- **Zero-setup onboarding** — start with the free model with no auth, then connect paid/account providers only when you need them

- **29 interface languages** - choose the app language and preferred agent communication language. Includes English, Chinese, Spanish, Hindi, Portuguese, French, Arabic, German, Japanese, Korean, Russian, Ukrainian, and 17 more.

- **Built-in code editor** — edit project files with Git support without leaving the app

- **Branch strategy** - choose per teammate at launch: use the main checkout or run selected agents in their own git worktree. You can still spell out branch rules in the provisioning prompt.

- **Team member stats** — global performance statistics per member

- **Attach code context** — reference files or snippets in messages, like in Cursor. You can also mention tasks using `#task-id`, or refer to another team with `@team-name` in your messages.

- **Notification system** — configurable alerts when tasks complete, agents need your response, new comments arrive, or errors occur

- **MCP integration** — supports the built-in `mcp-server` (see [mcp-server folder](./mcp-server)) for integrating external tools and extensible agent plugins out of the box

- **Post-compact context recovery** — when the active runtime compacts its context, the app restores the key team-management instructions so kanban/task-board coordination stays consistent and important operational context is not lost

- **Task context is preserved** — thanks to task descriptions, comments, and attachments, all essential information about each task remains available for ongoing work and future reference

- **Workflow history** — see the full timeline of each task: when and how its status changed, which agents were involved, and every action that led to the current state

</details>

## Comparison

| Feature                             | Agent Teams                                                                                             | Gas Town                                                             | Paperclip                                                              | Cursor                                                         | Claude Code CLI                                              | OpenRig |
| ----------------------------------- | ------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------- | ---------------------------------------------------------------------- | -------------------------------------------------------------- | ------------------------------------------------------------ | ------- |
| **Supported desktop platforms**     | macOS · Windows · Linux (native desktop app)                                                            | macOS · Windows · Linux (CLI binaries)                               | No native desktop app (self-hosted control plane)                        | macOS · Windows · Linux (desktop IDE)                         | macOS · Windows · Linux (CLI)                               | macOS · Linux; Windows via WSL2 (not native) |
| **Live team collaboration**         | ✅ █████████░ 9/10 · Agents plan, delegate, work, and review together                                   | ✅ ████████░░ 8/10 · Persistent teams with coordination and recovery | ⚠️ ███████░░░ 7/10 · Scheduled agents + MCP task collaboration, less live peer teamwork | ⚠️ ██████░░░░ 6/10 · Parallel agents + subagents, no peer team | ⚠️ ███████░░░ 7/10 · Experimental teams + cross-session messaging; recovery limits | ✅ ████████░░ 8/10 · Persistent roles, delegation, owned-work queues, and review handoffs |
| **Human control and approvals**     | ✅ Per-action approvals, roles, and notifications                                                       | ✅ Gates, roles, escalation, recovery                                | ✅ Board approvals, roles, pause, stop                                 | ⚠️ Command controls and admin policies                         | ✅ Permissions + hooks                                       | ✅ Runner permission requests, audited policies, typing guard, and stop controls |
| **Visual team progress**            | ✅ Teammates, tasks, blockers, handoffs, activity, logs                                                 | ⚠️ Agent tree + feed panels                                          | ⚠️ Org chart and run status, not a task/log map                        | ⚠️ Agents Window, no shared peer-team map                      | ⚠️ Terminal team panel, no graphical UI                      | ✅ TUI topology graph, seat state, missions, and activity feed |
| **Team workspace**                  | ✅ Tasks, code, terminal, review, and teammates in one app                                              | ⚠️ Mail/feed/dashboard across tools                                  | ⚠️ Board + task chats, less live teammate view                         | ⚠️ Agents Window, no peer team workspace                       | ❌ No desktop UI                                             | ⚠️ TUI, tmux, task-following terminal views; optional maintenance web UI |
| **Easy setup**                      | ✅ Start free, no signup or API key                                                                     | ❌ Manual CLI stack                                                  | ⚠️ `npx` onboarding + browser app                                      | ⚠️ App install + account                                       | ⚠️ CLI + env flag                                            | ⚠️ Node + tmux + provider login; guided setup on macOS/Linux |
| **Agent readiness and recovery**    | ✅ Clear ready/stuck status with launch diagnostics                                                     | ✅ Working/stalled/failed health with recovery controls              | ⚠️ Run status and orphan recovery                                      | ⚠️ Agent status in Agents Window                               | ⚠️ Statuses in terminal, no graphical UI                     | ✅ Readiness checks, snapshots, and explicit per-seat restore outcomes |
| **Kanban board**                    | ✅ 5 columns, real-time                                                                                 | ❌ Dashboard, not Kanban                                             | ✅ 7 columns, drag-and-drop                                            | ❌                                                             | ❌                                                           | ❌ Owned-work queue and mission views; no Kanban board |
| **Code review**                     | ✅ Agent review plus accept, reject, and comment in the app                                             | ⚠️ Merge queue, no diff UI                                           | ⚠️ Review gates + experimental GitHub review bot; no in-app hunk review                                | ✅ Local Agent Review + PR Bugbot                               | ⚠️ Agent review, no review UI                                | ⚠️ Reviewer rigs and proof approvals; no task-level hunk review UI |
| **Cross-team communication**        | ✅ Direct agent messages and shared task links across teams                                             | ⚠️ Mailboxes + handoffs                                              | ⚠️ Comments + @mentions                                                | N/A                                                            | ⚠️ Cross-session messaging, no shared cross-team task graph   | ✅ Cross-rig messages, broadcasts, chatrooms, and queue handoffs |
| **Linked tasks**                    | ✅ Tasks can link to and block each other                                                               | ✅ Task deps + grouped work                                          | ✅ Goals, parent tasks, blockers                                       | ❌                                                             | ✅ Shared task list                                          | ✅ Queue blockers, workflow dependencies, and transactional handoffs |
| **Agent activity and history**      | ✅ Messages, tool calls, timeline, token use, and cost                                                  | ⚠️ Session recall, feed, metrics                                     | ⚠️ Run transcripts + audited cost ledger (estimates possible)                                        | ⚠️ Agent chat + terminal                                       | ⚠️ CLI transcripts + background logs                         | ✅ Activity feed, transcripts, queue audit, and runtime-dependent telemetry |
| **Organizations & global overview** | ✅ Nested groups, live team/task status, relations, cross-team activity                                 | ⚠️ Coordination hierarchy, no editable org map                       | ✅ Org chart + board governance                                        | ⚠️ Team admin, no live org map                                 | ❌                                                           | ⚠️ Org/group policies + topology; no editable company org map |
| **Mixed AI teammates**              | ✅ Claude Code, Codex, OpenCode, Cursor, SuperGrok, GitHub Copilot, Z.AI, MiniMax, and Kiro in one team | ✅ Many providers, terminal-first                                    | ✅ Own runtimes + MCP clients; experimental Runner adapters                                      | ⚠️ Multi-model parent + subagents, no peer team                | ⚠️ Claude-only experimental teams                            | ✅ Claude, Codex, Pi, Oh My Pi + experimental hosted MCP members |
| **Budget controls**                 | ✅ Budget alerts + hard caps for scheduled runs                                                         | ⚠️ Cost tiers + digest, no hard caps                                 | ✅ Per-agent budgets + hard stops                                      | ⚠️ Usage + cloud spend limits                                  | ⚠️ Usage/org limits + print-mode hard cap                    | ⚠️ Token-burn/provider-window monitoring; no documented spend cap |
| **Separate agent workspaces**       | ✅ Optional workspace per teammate                                                                      | ✅ Core primitive                                                    | ✅ Worktrees / branches                                                | ✅ Agents Window worktrees                                     | ✅ Built-in for sessions and subagents                       | ⚠️ Per-seat working directories; externally managed Git worktrees |
| **Terminal**                        | ✅ Built-in visual terminal for team and local commands                                                 | ⚠️ Terminal-based workflow, no built-in terminal                     | ⚠️ Runs commands, no interactive terminal                              | ✅ Built-in IDE terminal                                       | ⚠️ Runs in your terminal                                     | ✅ tmux + herdr/cmux integration; optional maintenance web terminal |
| **Built-in code editor**            | ✅ With Git support                                                                                     | ❌                                                                   | ❌ Control plane, not editor                                           | ✅ Full IDE                                                    | ❌                                                           | ⚠️ Basic file editor in optional maintenance web UI |
| **Task attachments**                | ✅ Auto-attach, agents read & attach files                                                              | ❌ Not task-level                                                    | ✅ Docs, attachments, work products                                    | ⚠️ Prompt context, not task attachments                        | ⚠️ Prompt files/images, not task attachments                 | ⚠️ Queue evidence refs, proof media, and Slack file attachments |
| **Price**                           | **Free OSS UI + free model with no auth**, paid providers optional                                      | Free OSS, runtime plans needed                                       | Free OSS, self-hosted + infra                                          | Free + paid usage                                              | Claude plan or API usage                                     | Free OSS; Apache-2.0; provider usage and hosting costs apply |


_Live collaboration scores are qualitative editorial assessments of documented communication, task ownership and dependencies, completion, and review capabilities; they are not benchmark results._

<details>
<summary><strong>Sources and methodology</strong></summary>

<br />

Research refreshed November 6, 2026 using published releases and official docs: Agent Teams v2.18.6 (November 5), Gas Town v1.2.1, Paperclip v2026.1023.0, Claude Code v2.1.318, and OpenRig v0.7.8 (November 5). Desktop platform coverage is about native app/CLI support only; web and mobile are excluded. Experimental capabilities are explicitly qualified. Scores are editorial, not live benchmarks. [Updated evidence](docs/research/desktop-comparison-2026-11-06.md).

- **Agent Teams:** [collaboration controller](agent-teams-controller/src/internal), [organizations feature](src/features/organizations/README.md), [terminal workspace](src/features/terminal-workspace/renderer/ui/TerminalWorkspacePanel.tsx), [token usage budgets](src/features/token-usage/contracts/dto.ts), [scheduled run budget cap](src/main/services/schedule/ScheduledTaskExecutor.ts)
- **Gas Town:** [README](https://github.com/gastownhall/gastown), [provider guide](https://github.com/gastownhall/gastown/blob/main/docs/agent-provider-integration.md), [scheduler](https://github.com/gastownhall/gastown/blob/main/docs/design/scheduler.md), [dashboard source](https://github.com/gastownhall/gastown/blob/main/internal/web/templates/convoy.html), [v1.2.1 release](https://github.com/gastownhall/gastown/releases/tag/v1.2.1)
- **Paperclip:** [stable v2026.1023.0 release](https://github.com/paperclipai/paperclip/releases/tag/v2026.1023.0), [README](https://github.com/paperclipai/paperclip), [adapters](https://github.com/paperclipai/paperclip/blob/master/docs/adapters/overview.md), [heartbeat protocol](https://github.com/paperclipai/paperclip/blob/master/docs/guides/agent-developer/heartbeat-protocol.md), [org chart](https://github.com/paperclipai/paperclip#the-systems), [OrgChart source](https://github.com/paperclipai/paperclip/blob/master/ui/src/pages/OrgChart.tsx), [budgets](https://github.com/paperclipai/paperclip/blob/master/docs/guides/board-operator/costs-and-budgets.md), [runtime services](https://github.com/paperclipai/paperclip/blob/master/docs/guides/board-operator/execution-workspaces-and-runtime-services.md), [Kanban source](https://github.com/paperclipai/paperclip/blob/master/ui/src/components/KanbanBoard.tsx), [work products](https://github.com/paperclipai/paperclip/blob/master/packages/shared/src/validators/work-product.ts), [v2026.824.1 release (historical)](https://github.com/paperclipai/paperclip/releases/tag/v2026.824.1)
- **Cursor:** [desktop downloads](https://cursor.com/download), [terminal](https://cursor.com/docs/agent/tools/terminal), [Cloud Agents](https://cursor.com/docs/cloud-agent), [Agents Window](https://cursor.com/docs/agent/agents-window), [subagents](https://cursor.com/docs/subagents), [Agent Review](https://cursor.com/docs/agent/agent-review), [Bugbot](https://cursor.com/docs/bugbot), [worktrees](https://cursor.com/docs/configuration/worktrees), [models and pricing](https://cursor.com/docs/models-and-pricing), [team pricing](https://cursor.com/docs/account/teams/pricing)
- **Claude Code CLI:** [supported platforms](https://code.claude.com/docs/en/setup), [CLI](https://code.claude.com/docs/en/cli-usage), [agent teams](https://code.claude.com/docs/en/agent-teams), [cross-session messaging](https://code.claude.com/docs/en/cross-session-messaging), [worktrees](https://code.claude.com/docs/en/worktrees), [subagents](https://code.claude.com/docs/en/sub-agents), [dynamic workflows](https://code.claude.com/docs/en/workflows), [costs](https://code.claude.com/docs/en/costs), [Claude pricing](https://claude.com/pricing), [v2.1.246 release](https://github.com/anthropics/claude-code/releases/tag/v2.1.318)
- **OpenRig:** [stable v0.7.8 release](https://github.com/mvschwarz/openrig/releases/tag/v0.7.8), [platform requirements](https://github.com/mvschwarz/openrig/blob/main/docs/reference/getting-started.md), [historical v0.6.5 README](https://github.com/mvschwarz/openrig/blob/5ea35e93bca9460db94da0fc31afbc3716ea14ba/README.md), [OpenRig coordination and queue](https://github.com/mvschwarz/openrig/blob/5ea35e93bca9460db94da0fc31afbc3716ea14ba/docs/as-built/architecture/coordination-primitive.md), [OpenRig snapshot and restore](https://github.com/mvschwarz/openrig/blob/5ea35e93bca9460db94da0fc31afbc3716ea14ba/docs/as-built/architecture/lifecycle-snapshot-restore.md), [OpenRig workspace model](https://github.com/mvschwarz/openrig/blob/5ea35e93bca9460db94da0fc31afbc3716ea14ba/docs/as-built/architecture/workspace-primitive.md), [OpenRig basic web editor](https://github.com/mvschwarz/openrig/blob/5ea35e93bca9460db94da0fc31afbc3716ea14ba/packages/ui/src/components/files/FilesWorkspace.tsx); [row-by-row evidence](docs/research/openrig-comparison-2026-10-06.md)
- **Historical fact ledger (June–August 2026):** [comparison research notes](docs/research/gastown-paperclip-comparison-2026-06-25.md)

</details>

---

## Quick start

1. **Download** the app for your platform (see [Installation](#installation))
2. **Launch the desktop app** - start with the free model with no auth, or let the setup wizard detect runtimes and guide provider authentication
3. **Create a team** — Pick a project, define roles, write a provisioning prompt
4. **Watch** — Agents spawn, create tasks, and work. You see it all on the kanban board

The desktop app is the main product. The web version is currently in development.

---

## FAQ

<details>
<summary><strong>Why not just use Claude Code with pre-configured agents?</strong></summary>
<br />
We're seeing an interesting trend. First, AI agents started communicating with each other (something Claude and Codex didn't support before). Then we started seeing messenger-style AI assistants like Grok Bot, Meta Muse and OpenAI Dots, where you can create a team (or even multiple teams, as in Grok) and communicate with agents just like in a regular messenger. This is a much more natural and familiar way for people to interact with AI (like with humans).

For the general audience, that might be enough. But people often need to build many/large/complex  projects, and that's where bigger teams of agents become necessary, each with their own roles and responsibilities.

And when you have a lot of teammates, you need a good way to organize and manage them. That's exactly where Agent Teams comes in. You can see everything clearly in a visual interface, understand what's happening across your teams, and easily interact with individual agents.

Sure, you can set up something similar with regular Claude, but keeping track of everything without a proper frontend is difficult, and we are targeting precisely this multi-agent use case.

For us, it's not just about running multi-agent systems. It's about having full visibility into who's doing what, keeping everything organized, and making it easy to interact with your agents.
</details>

<details>
<summary><strong>Does it read or upload my code?</strong></summary>
<br />
The app reads local runtime/session data to power the UI, but it does not upload, sync, or send your project or code to Agent Teams servers. There is no Agent Teams cloud backend for storing your code. If you launch an external AI runtime, that runtime communicates directly with its provider under the provider's own terms. Setup checks only verify provider access and capabilities; they do not upload your project to Agent Teams.
</details>

<details>
<summary><strong>Can agents communicate with each other?</strong></summary>
<br />
Yes. Agents send direct messages, create shared tasks, and leave comments - all coordinated by the app's own orchestration layer.
</details>

<details>
<summary><strong>Is it free?</strong></summary>
<br />
Yes. The app is free and open source, and you can start with a free model with no auth - no registration, API keys, or credit card. If you want more models, connect the provider access you already have, including Claude Code, Codex, OpenCode/OpenRouter, Cursor, SuperGrok, GitHub Copilot, Z.AI, MiniMax, and Kiro.
</details>

<details>
<summary><strong>Can I review code changes before they're applied?</strong></summary>
<br />
Yes. Tasks with detected code changes include a full diff view where you can accept, reject, or comment on individual code hunks — similar to Cursor's review flow.
</details>

<details>
<summary><strong>What happens if an agent gets stuck?</strong></summary>
<br />
Send a direct message to course-correct, or open the teammate status to inspect diagnostics and restart the agent. Agent Teams also has a nudge system: the app can send a short control message when there is a clear reason to wake an agent up, such as after a known rate-limit cooldown, when a teammate has not synced with its current task or review, or when progress appears stalled. Nudges are guarded and rate limited, so they are meant to help the agent continue, not spam it. If an agent needs your input, you'll get a notification and the task will show a distinct badge on the board.
</details>

<details>
<summary><strong>Does it support multiple projects and teams?</strong></summary>
<br />
Yes. Run multiple teams in one project or across different projects, even simultaneously. To avoid Git conflicts, enable git worktree isolation for selected teammates when launching the team, and use the provisioning prompt for any extra branch or merge rules.
</details>

<details>
<summary><strong>What if the Linux app freezes or shows a blank window over RDP?</strong></summary>
<br />
Some RDP (Remote Desktop Protocol) sessions expose virtual GPU drivers that can break Electron rendering. Launch with `AGENT_TEAMS_DISABLE_GPU=1` to disable Electron hardware acceleration for that run, for example `AGENT_TEAMS_DISABLE_GPU=1 pnpm dev` from source or `AGENT_TEAMS_DISABLE_GPU=1 ./Agent.Teams.AI.AppImage` for AppImage builds.
</details>

---

## Roadmap (new)

- [ ] Launching 24/7 autonomous teams in the cloud (in progress)
- [ ] Automatic account switching (in progress)
- [ ] Navigating very complex long tasks without losing context (in progress)
- [x] Multiple AI runtimes and provider-backed coding agents (more in progress)
- [ ] A secure universal plugin system regardless of the type of agents ([in progress](https://github.com/777genius/universal-agent-plugins))
- [ ] Efficiency and controllability under load with 50-100+ agents in parallel (in progress)

## Vision

Agent Teams AI is moving toward a simple interface for highly autonomous work. From a phone, users should be able to see only the high-level tasks that matter and give the lead agent instructions by voice. The lead should route each request to the right project and team, while agents break it down, work in parallel, communicate, review, resolve blockers, and prepare the final result with minimal supervision.

Technical subtasks and coordination should remain available as a secondary layer, while the main view surfaces only the decisions, blockers, and outcomes that genuinely need attention. Agents should first try to solve problems together and involve the user only when human input is truly required.

This vision is not tied to a specific SDLC or workflow. Users should be free to organize teams however they prefer and continue using their existing plugins, skills, MCP servers, rules, and agent settings.

## Development

For feature architecture and implementation guidance:

- Canonical standard - [docs/FEATURE_ARCHITECTURE_STANDARD.md](docs/FEATURE_ARCHITECTURE_STANDARD.md)
- Repo working instructions - [CLAUDE.md](CLAUDE.md)
- Feature root guidance - [src/features/README.md](src/features/README.md)
- Reference implementation - `src/features/recent-projects`

## Tech stack

Electron 40, React 19, TypeScript 5, Tailwind CSS 3, Zustand 4. The desktop app reads local project, app, and runtime/session data. Claude Code data under `~/.claude/` is one supported local source, while some runtime modes may also use provider or startup capability services when required.

<details>
<summary><strong>Build from source</strong></summary>

<br />

**Prerequisites:** Node.js 26.12.1 Current (recommended for source builds) or Node.js 24.22.1 LTS, pnpm 11.22.0

On macOS, official Node.js 24 prebuilt binaries require macOS 13.5+.
Node.js 26 does not bundle Corepack; install the pinned pnpm version separately if `pnpm` is unavailable.

```bash
git clone https://github.com/777genius/agent-teams-ai.git
cd agent-teams-ai
pnpm install
pnpm dev
```

`pnpm dev` starts the desktop Electron app. Do not start a browser/web dev server for normal development; that path is limited and is not the supported way to run agent teams locally.

To run the desktop app against a local orchestrator checkout during development, point it at the
source launcher:

```bash
CLAUDE_AGENT_TEAMS_ORCHESTRATOR_CLI_PATH=/absolute/path/to/agent_teams_orchestrator/cli-source \
  pnpm dev
```

This runs the current orchestrator sources directly through Bun and avoids rebuilding the complete
runtime bundle on every app start. Do not use `CLAUDE_DEV_RUNTIME_ROOT` for the normal source
development loop: that override runs `bun run build:dev` before Electron starts and can make the app
appear stuck at `bun run ./scripts/build.ts --dev`. Use it only when intentionally validating the
generated development bundle.

Use `pnpm dev:mcp` for automated interactive or visual UI verification. It exposes the current
Electron renderer through the local Chrome DevTools Protocol endpoint on `127.0.0.1:9222`, avoiding
window-selection ambiguity when packaged or other Electron apps are also open.

The desktop app auto-discovers Claude Code projects from `~/.claude/`.

### Terminal Platform integration

Fresh clones install Terminal Platform SDK packages from `vendor/terminal-platform`, so
`pnpm install --frozen-lockfile` does not require a sibling `../terminal-platform` checkout.
Those artifacts are pinned in `vendor/terminal-platform/manifest.json` and can be refreshed from a
local Terminal Platform checkout with:

```bash
CLAUDE_TERMINAL_PLATFORM_ROOT=/path/to/terminal-platform pnpm terminal-platform:pack
```

For local Terminal Platform development, build Terminal Platform first, then run this app with the
same environment variable:

```bash
CLAUDE_TERMINAL_PLATFORM_ROOT=/path/to/terminal-platform pnpm dev:mcp
```

When the variable is set, the Electron build aliases Terminal Platform SDK imports to that local
checkout and the terminal runtime loads `terminal-platform-node` from the local native artifact.

### Debug teammate runtimes

Development launches use the app-managed process backend for teammates by default. To inspect
teammates in `tmux` panes while debugging, start the desktop app with:

```bash
CLAUDE_TEAM_TEAMMATE_MODE=tmux pnpm dev
```

The same override is available per launch from custom CLI args with
`--teammate-mode tmux`. Use this as an operator/debug mode; the default process backend provides
stronger app-owned lifecycle, diagnostics, and cleanup for normal team launches.

### Build for distribution

```bash
pnpm dist:mac:arm64  # macOS Apple Silicon (.dmg)
pnpm dist:mac:x64    # macOS Intel (.dmg)
pnpm dist:win        # Windows (.exe)
pnpm dist:linux      # Linux (AppImage/.deb/.rpm/.pacman)
pnpm dist            # Current platform
```

Distribution scripts run the production build and stage the bundled multimodel runtime from
`runtime.lock.json` before packaging. Use `pnpm clean:runtime` to remove staged runtime files after
local packaging.

### Scripts

| Command                       | Description                                                                  |
| ----------------------------- | ---------------------------------------------------------------------------- |
| `pnpm dev`                    | Desktop app development with hot reload                                      |
| `pnpm dev:mcp`                | Desktop app development with hot reload and local CDP debugging on port 9222 |
| `pnpm build`                  | Production build                                                             |
| `pnpm typecheck`              | TypeScript type checking                                                     |
| `pnpm lint`                   | Lint (no auto-fix)                                                           |
| `pnpm lint:fix`               | Lint and auto-fix                                                            |
| `pnpm format`                 | Format code with Prettier                                                    |
| `pnpm test`                   | Run all tests                                                                |
| `pnpm test:watch`             | Watch mode                                                                   |
| `pnpm test:coverage`          | Coverage report                                                              |
| `pnpm test:coverage:critical` | Critical path coverage                                                       |
| `pnpm check`                  | Full quality gate (types + lint + test + build)                              |
| `pnpm fix`                    | Lint fix + format                                                            |
| `pnpm quality`                | Full check + format check + knip                                             |

</details>

---
<!--
## Ideas

- [ ] Planning mode to organize agent plans before execution
- [ ] Visual workflow editor ([@xyflow/react](https://github.com/xyflow/xyflow)) for building and orchestrating agent pipelines with drag & drop
- [ ] Remote agent execution via SSH: launch and manage agent teams on remote machines over SSH (stream-json protocol over SSH channel, SFTP-based file monitoring for tasks/inboxes/config)
- [ ] Generic ACP runtime: connect Agent Client Protocol-compatible agents through one reusable runtime integration
- [x] Limited standalone dashboard for local or trusted-network use. It does not provide the full desktop runtime and should not be exposed directly to the internet without authentication and a reverse proxy.
- [ ] 2 modes: current (agent teams), and a new mode: regular subagents (no communication between them)
- [ ] Curate what context each agent sees (files, docs, MCP servers, skills)
- [ ] User-managed runtime profiles and built-in presets: choose a profile per team launch or schedule, isolate provider credentials and config roots (`CLAUDE_CONFIG_DIR`, `CODEX_HOME`, OpenCode XDG roots), optionally isolate the child-process `HOME`, and pin running teams to an immutable profile snapshot
- [x] Slash commands
- [ ] Outgoing message queue — queue user messages while the lead (or agent) is busy; clear agent-busy status in the UI; flush to stdin or relay from inbox when idle (durable queue on disk for the lead inbox path)
- [ ] `createTasksBatch` — IPC/service API to create many team tasks in one call (playbooks, markdown checklist import, scripts); complements single `createTask`
- [ ] Command palette — extend Cmd/Ctrl+K beyond project/session search to runnable actions (quick commands, navigation shortcuts, team/task operations) in a keyboard-first flow
- [ ] Custom kanban columns
- [x] Run terminal commands
- [x] Monitor agents processes/stats
- [ ] Reusable agents with SOUL.md
- [ ] Messenger integrations
- [ ] SDK to programmatically launch agents
      ...
-->
---

## Contributing

See [CONTRIBUTING.md](.github/CONTRIBUTING.md) for development guidelines. Please read our [Code of Conduct](.github/CODE_OF_CONDUCT.md).

By opening a pull request you agree to the [Contributor License Agreement](.github/CLA.md) v2. You keep copyright.

## Partnerships

We are open to partnerships and collaboration opportunities. If you see a way to create value together, we are ready to discuss mutually beneficial terms.

Contact: [iliya@agentteams.live](mailto:iliya@agentteams.live)

## Security

IPC and standalone HTTP handlers validate IDs, paths, and payload shape at the boundary. Project editing and write operations are constrained to the selected project root, while read-only discovery also accesses local Claude data under `~/.claude/` and app-owned state paths when required. Path traversal and sensitive config/credential targets are blocked. See [SECURITY.md](.github/SECURITY.md) for details.

The standalone HTTP dashboard is intended for local or trusted-network use and does not include built-in authentication for public deployment. Do not expose it directly to the internet without an authenticated reverse proxy.

GitHub Dependabot helps surface dependencies with known vulnerabilities and available security updates.

## License

[AGPL-3.0](LICENSE)

Opening a pull request agrees to [CLA v2](.github/CLA.md). That does not replace AGPL-3.0 on the public repository. It lets the project owner sublicense that work under additional terms (for example a commercial license) while keeping the AGPL-3.0 grant from the submission date.

桌面应用安装包（macOS/Windows/Linux，见 Releases）；可选连接本地已安装的 Claude Code、Codex、OpenCode 运行时；可连接 Cursor、SuperGrok、GitHub Copilot、Z.AI、MiniMax、Kiro 等账号或 API Key；免费模型的提供方、额度与费用待核验；README 未说明 Node/开发环境前置要求；MCP 服务端方案待核验；国内网络与合规可用性待核验。