ergo-agent-pay (Python)

Python reference SDK for the Ergo rail of Accord Protocol. Agent payments, Note operations, balance + UTxO inspection, LangChain / OpenAI integrations.

The package keeps its legacy npm-mirror name. A canonical Python Accord layer (port of @accord-protocol/core) is not yet published; when it is, it will live under a different PyPI name and import from this package as a rail dependency.

Zero required dependencies — uses Python standard library only.

pip install ergo-agent-pay

Quick start

from ergo_agent_pay import ErgoAgentPay

agent = ErgoAgentPay(address="YOUR_ADDRESS", network="testnet")

# Check balance
balance = agent.get_balance()
print(f"Balance: {balance['ergs']} ERG")

# Inspect a Note
note = agent.check_note("abc123...")
print(f"Note value: {note.value_erg} ERG")
print(f"Expired: {note.is_expired}")
print(f"Task hash: {note.task_hash}")

# Compute task hash for acceptance predicate
task_hash = ErgoAgentPay.compute_task_hash("task output here")

LangChain integration

from ergo_agent_pay import ErgoAgentPay

agent = ErgoAgentPay(address="YOUR_ADDRESS", network="testnet")
tool = agent.as_langchain_tool(server_url="http://localhost:3000")

# Use in any LangChain agent
from langchain.agents import AgentExecutor
tools = [tool]

OpenAI function calling

definition = agent.as_openai_function()

response = openai.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Pay 0.001 ERG to <address>"}],
    tools=[{"type": "function", "function": definition}],
)

Transaction building

Python doesn't have a Fleet SDK equivalent. For full transaction building, use:

• TypeScript SDK (recommended): run ergo-agent-pay TypeScript server, call via HTTP
• ergpy: pip install ergpy — AppKit JVM wrapper
• sigma-rust: Python bindings for Rust-based TX builder

See example 06 for the HTTP delegation pattern.

Note lifecycle — what Python can do

API reference

Full API reference: docs/api-reference.md