Build and manage a team of AI agents from a single dashboard.
TeamHero is an open-source, local-first multi-agent orchestration platform powered by Claude CLI. Create specialized AI agents, assign tasks, review deliverables, and coordinate work — all from a web dashboard running on your machine.
No cloud. No subscriptions. Your data stays local.
| TeamHero | CrewAI | LangGraph | AutoGen | |
|---|---|---|---|---|
| License | MIT | MIT | MIT | MIT |
| Local-first | Yes | Cloud-focused | Cloud-focused | Local |
| Dashboard | Built-in web UI | Paid (cloud) | None | None |
| Setup | npm install + go |
Python + config | Python + config | Python + config |
| Agent memory | Built-in per-agent | Cloud only | Manual | Manual |
| Task management | Built-in lifecycle | API only | None | None |
| Knowledge base | Built-in | None | None | None |
| Price | Free | Free tier, then $99/mo+ | Free tier, then paid | Free |
TeamHero gives you a complete agent management platform out of the box — dashboard, task tracking, knowledge base, media library, and round table reviews — with zero configuration.
- Node.js 18+
- Claude CLI — the AI backbone that powers your agents
npm install -g @anthropic-ai/claude-codegit clone https://github.com/sagiyaacoby/TeamHero.git my-team
cd my-team
npm installWindows:
launch.batMac / Linux:
bash launch.shThe dashboard opens at http://localhost:3777. That's it.
- Complete the setup wizard — set your name, role, and goals
- Go to the Command Center and ask the orchestrator to build a team
"Build me a team with a Content Writer, a Researcher, and a Developer"
- Create tasks and assign them to agents
- Review deliverables and approve or request revisions
- Promote finished work to the Knowledge Base
Web-based portal to manage your entire agent team. View agent status, tasks, and deliverables at a glance.
Talk directly to your orchestrator agent via an integrated terminal. The orchestrator coordinates all other agents.
Full task lifecycle with version tracking:
draft -> in_progress -> pending_approval -> approved / revision_needed -> done
Each task tracks versions, deliverables, attachments, and owner feedback.
Every agent has its own name, role, personality traits, tone, rules, and persistent memory. Agents remember context across sessions.
Promote task deliverables into a searchable, categorized knowledge library. Tag, filter, and browse research and outputs.
Shared media storage with thumbnail previews. Upload images, documents, and files for agents to reference.
Structured review sessions across all agents. Get a status report, surface blockers, and make decisions.
Enable optional capabilities like browser automation (Playwright) and screen recording.
TeamHero is intentionally simple:
- Server: Single
server.jsfile using Node's built-inhttpmodule. No Express, no frameworks. - Portal: Vanilla HTML, CSS, and JavaScript. No React, no build step, no bundlers.
- AI: Claude CLI runs as a subprocess. Agents are Claude sessions with custom system prompts.
- Data: JSON files on disk. No database required.
TeamHero/
├── server.js # API server
├── portal/ # Dashboard (vanilla HTML/CSS/JS)
│ ├── index.html
│ ├── css/style.css
│ └── js/app.js
├── agents/ # Agent definitions and memory
├── config/ # Team rules, security rules
├── data/
│ ├── tasks/ # Tasks with version folders
│ ├── knowledge/ # Knowledge base documents
│ ├── round-tables/ # Round table summaries
│ └── media/ # Shared media library
├── profile/ # Owner profile
├── launch.bat # Windows launcher
└── launch.sh # Mac/Linux launcher
The server exposes a REST API at http://localhost:3782:
| Endpoint | Method | Description |
|---|---|---|
/api/agents |
GET | List all agents |
/api/agents |
POST | Create a new agent |
/api/agents/:id |
PUT | Update an agent |
/api/tasks |
GET | List all tasks |
/api/tasks |
POST | Create a task |
/api/tasks/:id |
PUT | Update a task |
/api/tasks/:id/versions |
GET | Get task version history |
/api/tasks/:id/promote |
POST | Promote deliverable to Knowledge Base |
/api/knowledge |
GET | List knowledge documents |
/api/knowledge |
POST | Create a knowledge document |
/api/profile |
PUT | Update owner profile |
Contributions are welcome! See CONTRIBUTING.md for guidelines.
Whether it's a bug fix, new feature, documentation improvement, or agent template — we'd love your help.
| Problem | Solution |
|---|---|
| Port already in use | Dashboard uses 3777, API uses 3782. Free these ports or stop conflicting processes. |
| Claude CLI not found | Run npm install -g @anthropic-ai/claude-code and ensure it's on your PATH. |
| npm install fails | Make sure you have Node.js 18+. Check with node -v. |
| Dashboard won't load | Check that server.js is running. Look at the terminal for errors. |
- Agent marketplace (community templates and skill packs)
- Advanced analytics and agent performance metrics
- Vector DB integration for long-term agent memory
- Multi-user collaboration
-
npx create-teamheroone-command setup
MIT - Created by Sagi Yaacoby