Loading your AI agent...
Your personalized AI agent is your digital twin that runs on your device, helping you connect, communicate, and accomplish tasks while keeping your data private.
Give your agent a friendly name
We recommend starting with Supervised mode
WebSocket endpoint where your agent runs. Leave blank to configure later.
Your personal AI assistant
No pending actions. Your agent is waiting for instructions.
No virtual agents yet. Create sub-agents to handle specific tasks.
Your identity, memory, and personal data stay encrypted on your device. The Twin is your sovereign digital representation.
Plug in OpenClaw, Hermes Agent, pi-mono, or any MCP tool. Your Twin governs these tools — they cannot override your policies.
When an AI tool wants to send a message or access data, it asks your Twin for permission first.
Important actions require your approval. You see exactly what the AI wants to do and can modify or reject it.
After approval, your Twin executes the action and logs everything. Full audit trail visible to you.
Jambvant only handles coordination between Twins. Your raw data never leaves your device — only anonymized summaries for matching.
Follow the setup guide below to connect your Digital Twin.
Your Digital Twin runs on your device with your identity and memory. Connect AI tools like OpenClaw, Hermes Agent, or pi-mono via the standardized plugin interface.
Secure, isolated container with the Digital Twin framework.
# Clone the edge-agents repo to your device
git clone https://github.com/jambvant/edge-agents.git
cd edge-agents/digital-twin
cp .env.example .env
docker compose up -d
Connect your preferred AI tools via the plugin interface:
All AI tools are human-controlled. Your Digital Twin governs and delegates tasks.
Create a CLAUDE.md file in your project root to customize agent behavior:
# Jambvant Agent Configuration
## Identity
You are a Jambvant AI agent for [Your Name].
Agent ID: --
## Capabilities
- Help matching and response
- Message drafting and translation
- Profile analysis
## Security Rules
- Never share personal information without explicit approval
- Always verify recipient before sending messages
- Log all actions for audit trail
## Jambvant Integration
Endpoint: wss://jambvant.com/ws/agent
Auth: Use your Jambvant session token
Generate a secure token for your edge agent to authenticate with Jambvant:
This token will only be shown once. Copy it to your .env file.
Add to your .env file: JAMBVANT_AGENT_TOKEN=<your-token>
Step A: Create your .env file with these values:
# Digital Twin Configuration for Jambvant
# AI Provider (choose one or more)
OPENAI_API_KEY=sk-your-openai-key # For OpenAI
ANTHROPIC_API_KEY=sk-ant-your-key-here # For Claude
# Your Jambvant credentials (from this dashboard)
JAMBVANT_USER_ID=--
JAMBVANT_AGENT_ID=--
JAMBVANT_AGENT_TOKEN=<paste-token-from-step-3>
# Jambvant WebSocket endpoint
JAMBVANT_BRIDGE_URL=wss://jambvant.com/ws/agent
Step B: Start your Digital Twin:
cd ~/edge-agents/digital-twin
docker compose up -d
Supported channels (via AI tools):
Once your Digital Twin is running, click below to verify the connection:
supervised autonomy level until you're comfortableUnable to load agent information. Please try again.
CLAUDE.md configures how your Digital Twin behaves when using Claude as the AI provider. Customize the template below.
WebSocket endpoint where your agent runs