Revertly AI - Git for AI Agents

This project is a demo for Revertly AI, an AI agent management platform that provides cross-system, dependency-aware rollback capabilities. Think of it as "Git for AI Agents" - giving you the undo button every company needs before deploying AI agents.

Features

• 🤖 AI Agent Management - Track and manage multiple AI agents across your organization
• 🔄 Cross-System Rollback - Revert agent actions across Confluence, GitHub, and more with one click
• 📊 Real-time Timeline - Monitor all agent actions in real-time with complete audit trails
• 🔍 Action Details - View before/after states for every change
• ✅ Service Integration - Connect agents to Confluence, GitHub, and other services
• 🎯 Demo Mode - Simulate agent actions for demonstrations and testing

Quick Demo

Want to see it in action? Check out the DEMO_GUIDE.md for a complete walkthrough of the demo flow.

Prerequisites

• Node.js and npm
• Python 3.10

Frontend (UI)

Setup and Run

1. Navigate to the UI directory:

   cd web

2. Install dependencies:

   npm install

3. Run the application:

   npm start

The UI will be available at http://localhost:3000.

Backend (Legacy)

Setup and Run

1. Create and activate a virtual environment:

   • Create the virtual environment:

     python3.10 -m venv venv

   • Activate the virtual environment:
     • Windows:

       venv\Scripts\activate

     • macOS/Linux:

       source venv/bin/activate

2. Navigate to the backend directory:

   cd src

3. Install dependencies:

   pip install -r requirements.txt

4. Run the application:

   uvicorn main:app --reload

The backend API will be available at http://127.0.0.1:8000.

MCP Server (Confluence Integration)

This server integrates with Confluence Cloud to track and revert AI agent actions.

Setup and Run

1. Create and activate a virtual environment:

   • If you haven't already, create a virtual environment in the project root:

     python3.10 -m venv venv

   • Activate the virtual environment:
     • Windows:

       venv\Scripts\activate

     • macOS/Linux:

       source venv/bin/activate

2. Navigate to the MCP server directory:

   cd mcp

3. Install dependencies:

   pip install -r requirements.txt

4. Run the application:

   uvicorn main:app --reload

The MCP server API will be available at http://127.0.0.1:8000.

Registering Agents

Before an AI agent can interact with Confluence via the MCP server, it needs to be registered. Use the POST /api/agents/register endpoint (documented in mcp/API_ENDPOINTS.md) to register your agents, providing their Confluence user email and API token.

Note: A demo agent is pre-registered for testing purposes.

Testing the Demo

Quick Test

Run the test script to verify everything is working:

python test_demo.py

This will test all the API endpoints and simulate an agent action.

Manual Testing

1. Start both the backend and frontend (see setup instructions above)
2. Open your browser to http://localhost:3000
3. Click "Simulate Agent Action" in the demo controls
4. View the action in the timeline below
5. Click "Details" to see the full change information
6. Click "Revert" to test the rollback functionality

Demo Flow

For a complete demo script and presentation guide, see DEMO_GUIDE.md.

The demo showcases:

1. Dashboard with AI agents and connected services
2. Real-time timeline of agent actions
3. Detailed action information with before/after states
4. One-click revert functionality
5. Visual feedback and notifications

Architecture

┌─────────────────┐
│  React Frontend │  (Port 3000)
│  - Dashboard    │
│  - Timeline     │
│  - Agent Cards  │
└────────┬────────┘
         │
         │ HTTP/REST
         │
┌────────▼────────┐
│   MCP Server    │  (Port 8000)
│   - FastAPI     │
│   - Agent Mgmt  │
│   - Change Track│
└────────┬────────┘
         │
         │ API Calls
         │
┌────────▼────────┐
│   Confluence    │
│     GitHub      │
│   (Services)    │
└─────────────────┘

API Documentation

See mcp/API_ENDPOINTS.md for complete API documentation.

Key endpoints:

• GET /api/agents - List all agents
• GET /api/changes - Get timeline of changes
• POST /api/demo/simulate_action - Simulate an agent action (demo mode)
• POST /api/confluence/revert_page/{change_id} - Revert a change

Customization

Branding

• Update theme colors in web/src/App.js
• Modify the sidebar in web/src/SideNav.js
• Change the logo and company name

Adding Services

• Update agent service connections in web/src/Agents.js
• Add new API integrations in the mcp directory
• Implement service-specific rollback logic

Real Confluence Integration

1. Update secrets/confluence.json with your credentials
2. Register agents with real Confluence API tokens
3. Use the actual Confluence API endpoints instead of demo mode

Troubleshooting

Port Already in Use

If port 8000 or 3000 is already in use:

• Backend: uvicorn main:app --reload --port 8001
• Frontend: Set PORT=3001 before running npm start

CORS Errors

Update the origins list in mcp/main.py to include your frontend URL.

Dependencies Issues

• Make sure you're using Python 3.10
• Delete node_modules and run npm install again
• Recreate the virtual environment if needed

Contributing

This is a demo project for Revertly AI. For production use or pilot program inquiries, contact the team.

License

Proprietary - Revertly AI