# AgenticSquad.dev

> AgenticSquad.dev gives your AI coding assistant a team: a Scrum Master plus a Tech Lead, Developer and QA for each platform (Web, Android, iOS). You talk to the Scrum Master, it puts the work on a Kanban board as cards, and the other agents pick the cards up and finish them. A local portal shows every handoff live on an office floor plan and on the board. Free, MIT licensed, runs on your machine.

- Website: https://agenticsquad.dev/ (Arabic: https://agenticsquad.dev/ar/)
- npm package: `agenticsquad` (https://www.npmjs.com/package/agenticsquad)
- Current version: 1.3.0
- Licence: MIT
- Author: Syed Ahmad Shahwaiz (https://www.linkedin.com/in/ahmadshahwaiz/)

## Install

Needs Node.js. In a terminal:

```bash
cd your-project
npx agenticsquad
```

Nothing is installed globally and setup takes under 5 seconds. The portal opens at http://localhost:3000 (or the next free port). If the project has no team yet, a setup screen lets you pick a Web, Android or iOS team, or all three.

Options: `npx agenticsquad --port 3001` for another port, `npx agenticsquad --dir /path/to/project` to watch another folder. Several projects can run at once on different ports.

## Supported AI coding tools

- Claude Code: reads `CLAUDE.md` and subagents in `.claude/agents/`
- OpenAI Codex: reads `AGENTS.md` and `.codex/`
- Google Gemini and Antigravity: read `GEMINI.md`
- OpenCode, including local models: reads `.agents/registry.yaml` and `.agents/definitions/*.md`

The team is defined once as Markdown and YAML in `.agents/`. `npx -p agenticsquad agents sync` writes it out for every tool. `npx -p agenticsquad agents validate` and `npx -p agenticsquad agents doctor` check the setup.

## How a request moves

1. **Request.** You describe the goal in your usual AI tool. That session only dispatches: it hands the request to the Scrum Master.
2. **Plan.** The Scrum Master sizes the request and picks the platform team. A small change (one to three files, no change to a shared type, API or stored data) goes straight to that platform's Developer with the Tech Lead as reviewer. Anything bigger becomes a card for the Tech Lead.
3. **Contract.** The Tech Lead writes a short contract for the Developer: files to change, files affected, risks and how to verify. If real testing is needed it opens a QA card.
4. **Build.** The Developer implements one card at a time, checks every acceptance criterion and moves the card to review.
5. **Test and review.** QA records pass or fail per criterion; the Tech Lead reviews against the contract. When every task is done the requirement closes.

Every brief, decision and result lives on the cards, so each agent reads its card instead of a long pasted prompt.

## What the portal shows

- Office floor plan: agents walk cards to each other on handoffs.
- Kanban board: Backlog, In Progress, Review, Done, with timestamps and assignees.
- Handoff stream: every delegation with the exact brief passed.
- Tool analytics: terminal commands, MCP tools, handbacks and file writes per agent.
- Code change inspector: the diff each agent produced, filed under its card.
- Time and tokens: work time and token usage per role.
- Replay: plays a finished requirement's handoffs on the floor plan; can be saved as video.
- Watch demo: a scripted walkthrough through your own team that saves nothing.

## Privacy

The portal listens only on 127.0.0.1 and refuses requests from other websites. It reads each tool's session history from your own disk (for Claude Code, `~/.claude/projects/`). Code, file paths, card text and tokens are never sent anywhere. The portal sends anonymous usage statistics (Google Analytics); turn them off in Settings or start with `VO_ANALYTICS=off npx agenticsquad`.

## FAQ

**Is it free?** Yes, free and MIT licensed.

**Can I change the agents?** Yes. Edit them in Team Studio inside the portal or in `.agents/definitions/`, then run `npx -p agenticsquad agents sync`. Roles, names, allowed tools, skills, rules and avatars are all configurable.

**Does my code leave my machine?** No. Everything runs locally; only anonymous usage statistics are sent, and they can be switched off.

## Pages

- [Home](https://agenticsquad.dev/)
- [Home in Arabic / الصفحة الرئيسية](https://agenticsquad.dev/ar/)
- [How to run a multi-agent team in Claude Code](https://agenticsquad.dev/claude-code-multi-agent-team/)
- [Sitemap](https://agenticsquad.dev/sitemap/)

## ملخص بالعربية

يمنح AgenticSquad.dev مساعد البرمجة بالذكاء الاصطناعي فريقًا كاملًا: سكرم ماستر، ومعه قائد تقني ومطوّر ومختبِر جودة لكل منصة (الويب وأندرويد وiOS). للتثبيت: انتقل إلى مجلد مشروعك باستخدام `cd` ثم شغّل `npx agenticsquad`. يعمل مع Claude Code وCodex وGemini وOpenCode، ويعمل محليًا على جهازك، وهو مجاني برخصة MIT.