radius-agent-template

A minimal NEST agent that can send and receive payments on the Radius network. Fork this, set your env vars, and you have a payment-capable agent.

Quick Start

# Clone
git clone https://github.com/radius-workshop/radius-agent-template.git
cd radius-agent-template

# Install
pip install -r requirements.txt

# Configure
cp .env.example .env
# Edit .env: add RADIUS_PRIVATE_KEY (required), ANTHROPIC_API_KEY (optional)

# Run
python agent.py

The agent starts on http://localhost:6000/a2a and registers with the NANDA agent network.

Two Modes

Command Routing (no LLM)

If ANTHROPIC_API_KEY is not set, the agent uses simple command routing:

balance              → Check your RUSD + SBC balances
balance 0x1234...    → Check someone else's balances
send 0x1234... 1.5   → Send 1.5 SBC to an address
status 0xabcd...     → Check a transaction's status
faucet               → Request testnet SBC
info                 → Chain info (ID, block, gas price)
help                 → List commands

Natural Language (with LLM)

If ANTHROPIC_API_KEY is set, the agent uses Claude with tool-calling. Users can ask things like:

• "What's my balance?"
• "Send 0.5 SBC to 0x1234..."
• "Did my last transaction go through? Here's the hash: 0xabcd..."
• "Get me some testnet tokens"

Files

File	Purpose
agent.py	Entry point — NANDA setup, command routing, LLM handler
tools.py	Tool definitions (Anthropic schema) + execution
radius_wallet.py	Vendored wallet library (balance, send, faucet, contracts)
nest.yaml	NEST agent configuration
.env.example	Environment variable template
Dockerfile	Container deployment (Railway, etc.)

How It Works

1. NANDA registration — The agent registers with the NANDA agent network on startup, making it discoverable by other agents via the A2A protocol.

2. Message handling — Incoming messages (from other agents or users) are routed to agent_logic(), which either parses commands directly or sends them to Claude for tool-calling.

3. Wallet operations — The radius_wallet.py module handles all blockchain interaction: signing transactions locally with a private key, querying the Radius RPC, and interacting with the faucet.

4. No external signer — Unlike the Wallet Concierge which uses a separate Privy signer service, this template signs locally. This is simpler but only suitable for testnet/hackathons.

Talking to This Agent

From another NEST agent or any A2A client:

import httpx

response = httpx.post("http://localhost:6000/a2a", json={
    "jsonrpc": "2.0",
    "method": "message/send",
    "id": "1",
    "params": {
        "message": {
            "parts": [{"kind": "text", "text": "What's my balance?"}],
            "messageId": "msg-1",
            "role": "user",
        }
    }
})
print(response.json())

Deploy to Railway

1. Fork this repo
2. Create a new Railway project
3. Set environment variables (RADIUS_PRIVATE_KEY, optionally ANTHROPIC_API_KEY)
4. Deploy — Railway will build from the Dockerfile

Alternatives

• Hermes Agent Template — A different approach using the Hermes framework (Python + Bun). More features (Telegram, Linear, DID-Web) but more complex.
• Wallet Concierge — Full-featured NEST agent with LangGraph, Privy signing, and Telegram bridge. Production-oriented but heavier.

Production Notes

This template uses a local private key for signing. For production:

1. Switch to Privy embedded wallets for secure key management
2. Run the signer as a separate service (see Wallet Concierge architecture)
3. Set NANDA_WALLET_PROVIDERS=privy and configure Privy env vars