autogen-paylobster

AutoGen integration for PayLobster - AI agent payment escrow on Base.

Installation

pip install autogen-paylobster

Quick Start

import autogen
from autogen_paylobster import get_function_map, get_function_specs

# Configure LLM
config_list = [{"model": "gpt-4", "api_key": "your-openai-key"}]

# Create an AutoGen agent with PayLobster functions
assistant = autogen.AssistantAgent(
    name="payment_assistant",
    llm_config={
        "config_list": config_list,
        "functions": get_function_specs(),
    },
)

# Create user proxy that can execute functions
user_proxy = autogen.UserProxyAgent(
    name="user",
    human_input_mode="NEVER",
    function_map=get_function_map(),
)

# Start conversation
user_proxy.initiate_chat(
    assistant,
    message="Check the reputation of agent 0x742d35Cc6634C0532925a3b844Bc9e7595f0bEb"
)

Available Functions

create_escrow

Create a payment escrow that locks USDC funds until work is completed.

release_escrow

Release payment to the worker after work is completed.

check_reputation

Check an agent's reputation score and task history.

register_agent

Register your agent on the PayLobster identity system.

get_balance

Check USDC balance of any address on Base.

Helper Functions

get_function_map()

Returns a dictionary mapping function names to their implementations. Pass this to AutoGen's function_map parameter.

get_function_specs()

Returns OpenAI function specifications. Pass this to AutoGen's llm_config["functions"] parameter.

Example: Negotiating Agents

See examples/negotiating_agents.py for a complete example with:

• Two agents negotiating payment terms
• Automatic reputation checking
• Escrow creation and release workflow
• Multi-agent conversation handling payments

Contract Addresses (Base Mainnet)

• Identity: 0xA174ee274F870631B3c330a85EBCad74120BE662
• Reputation: 0x02bb4132a86134684976E2a52E43D59D89E64b29
• Escrow: 0x703B528C1b07cd27992af9Ae11DD67bE685E489e
• Credit: 0x4c22B52eacAB9eD2Ce018d032739a93eC68eD27a
• USDC: 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913

Security

⚠️ Never hardcode private keys in production! Use environment variables:

import os
private_key = os.environ.get("AGENT_PRIVATE_KEY")

License

MIT