Ergo Agent Economy Q2-2026: Quarterly progress report

Q2 2026 · Quarterly Update

ergo-agent-pay v0.2

Q2 2026 Update: What Shipped in Ergo's Agent Economy Stack

Q: Which AI agent frameworks now have native Ergo payment support?

LangChain (via asLangChainTool() adapter), OpenAI function calling (via asOpenAIFunction() adapter), CrewAI, and AutoGen all have working examples in the ergo-agent-economy repo. There is also a Python SDK (ergo-agent-py) for non-Node agents and an MCP server (ergo-agent-mcp) so any MCP-compatible client (Claude Desktop, Cursor, etc.) can pay for things on Ergo testnet.

Q: Is the agent payment stack production-ready?

The Ergo protocol primitives (eUTXO, ErgoScript, Babel Fees, Sigma Protocols) are live on mainnet since 2019. The ergo-agent-pay SDK and ChainCash reference implementation are open-source prototypes — no security audits or production releases yet. Treat them as pre-production code, suitable for testnet development and early mainnet experiments with small amounts.

Q: Where can I clone and run the examples?

The full repo is at https://github.com/bez111/ergo-agent-economy. Examples are in the examples/ directory: 01-basic-payment, 02-note-payment, 03-acceptance-predicate, 04-orchestrator-budget, 05-api-payment-server, 06-python-agent, 07-streaming-pay, 08-treasury-multisig, 09-crewai-agents, 10-autogen-agent. Each has a self-contained README and runs against Ergo testnet — no mainnet ERG required.

Q: What is the Basis layer mentioned in the repo?

Basis is the off-chain credit and IOU layer built on top of Ergo's on-chain primitives. Off-chain Notes circulate cheaply between trusted parties; on-chain Reserves act as collateral and global settlement when trust isn't enough. This combines low-cost rapid payments with the option to settle on-chain via Reserve redemption. Basis is documented in the repo's docs/basis/ directory.

Two months after the first release. Three SDK packages, ten working examples, four AI framework adapters, an MCP server, and the full Note lifecycle. Here's what changed — and what we'd like to ship next.

EBy Ergo Developer Relations · Developer Advocate · Ergo PlatformMay 6, 20268 min read

View Repo

TL;DR

v0.2 — Full Note Lifecycle

Reserve creation, Note redemption, Tracker deployment, batch settlement. The full Reserve → Note → Tracker → Acceptance Predicate pipeline ships in one SDK.

Four AI Framework Adapters

LangChain, OpenAI function calling, CrewAI, AutoGen. Working examples in the repo. MCP server for Claude Desktop / Cursor / any MCP-compatible client.

Three Languages — One Stack

ergo-agent-pay (TypeScript), ergo-agent-py (Python), ergo-agent-mcp (MCP server). Same primitives, different runtimes.

Basis Layer Documented

Off-chain IOUs + on-chain settlement. Combines low-cost rapid payments with on-chain finality when trust isn't enough.

What Shipped Since Launch

The ergo-agent-pay SDK launched on 2026-03-18 with three examples and the basic pay() / issueNote() API. Three days later, v0.2 added the full Note lifecycle. Since then:

2026-03-18v0.1.0

Initial release: ErgoAgentPay class, policy engine, LangChain + OpenAI adapters, acceptance predicate helpers, NetworkClient. Examples 01-03.

2026-03-21v0.2.0

Full Note lifecycle: createReserve(), redeemNote(), settleBatch(), deployTracker(). Examples 04-06 (orchestrator, API server, Python agent). Register decode helpers. New error codes.

2026-04 — early MayEcosystem build-out

ergo-agent-py (Python SDK). ergo-agent-mcp (MCP server). Examples 07-10 covering streaming pay, treasury multisig, CrewAI, AutoGen. Basis off-chain credit layer documented.

Three Packages, One Repo

Same primitives — Reserve, Note, Tracker, Acceptance Predicate — exposed in TypeScript, Python, and over the Model Context Protocol.

ergo-agent-pay

TypeScript / Node.js

Primary SDK. ErgoAgentPay class, policy engine, LangChain + OpenAI adapters, full Note lifecycle methods. NPM-installable.

$ npm install ergo-agent-pay

ergo-agent-py

Python

Python equivalent for non-Node agents. Works with LangChain Python, CrewAI, AutoGen, and any Python-native AI agent framework.

$ pip install ergo-agent-py

ergo-agent-mcp

MCP Server

Model Context Protocol server. Plug into Claude Desktop, Cursor, or any MCP client — your AI assistant can now pay for things on Ergo testnet.

$ npx ergo-agent-mcp

10 Working Examples

Each example is self-contained and runs against Ergo testnet. No mainnet ERG required — the testnet faucet covers everything.

Basic Payment

Send ERG, agent → agent. The hello-world of on-chain agent payments.

Note Payment

Issue a programmable Note instead of raw ERG. Redeem it later.

Acceptance Predicate

Embed a task-completion condition in the Note. Receiver only redeems if condition holds.

Orchestrator Budget

Parent agent issues budgeted Notes to 3 sub-agents with per-agent acceptance predicates.

API Payment Server

Express server verifies Note on-chain before serving the response. Includes client demo.

Python Agent

Python/LangChain agent that pays for API calls using Ergo Notes.

Streaming Pay

Continuous micropayments for streaming services (per-token billing for LLM inference).

Treasury Multisig

Multi-agent treasury controlled by m-of-n signatures. Spending policy enforced on-chain.

CrewAI Agents

Multi-agent crew where each agent has its own wallet and pays peers for sub-tasks.

AutoGen Agent

Microsoft AutoGen agent with Ergo payment tool registered for autonomous transactions.

Agent Framework Adapters

Drop-in integrations for the most common AI agent frameworks. Each one is one import + one line of registration.

LangChain

Drop-in tool that exposes pay(), issueNote(), checkNote() to any LangChain agent. Works with both JS and Python LangChain.

const tool = agent.asLangChainTool()
// Add to LangChain agent's toolkit

OpenAI Function Calling

Returns a function definition compatible with OpenAI's tool-use API. Your GPT-4-class model can call pay() directly.

const fn = agent.asOpenAIFunction()
// Register with chat.completions.create({ tools: [fn] })

CrewAI

Each CrewAI agent gets its own wallet and policy. Agents pay each other for sub-task delivery.

# example 09-crewai-agents
from ergo_agent_py import ErgoAgent
agent = ErgoAgent(...)

AutoGen

Microsoft AutoGen agent with Ergo payment registered as a callable tool. End-to-end agent-to-agent settlement.

# example 10-autogen-agent
autogen_agent.register_tool(ergo_pay_tool)

MCP (Claude / Cursor / etc.)

Model Context Protocol server. Any MCP-compatible client (Claude Desktop, Cursor, Zed, etc.) can pay on Ergo.

# Add to MCP config:
{ "ergo": { "command": "npx ergo-agent-mcp" } }

Full Note Lifecycle

v0.2 covers the complete Reserve → Note → Tracker → Acceptance Predicate pipeline. These are the methods you call to issue, redeem, and settle programmable bearer instruments.

createReserve()

Deploy a Reserve collateral box (P2PK for dev, custom ErgoScript for production).

issueNote()

Mint a programmable Note from an existing Reserve. Encodes value, expiry, and acceptance predicate.

checkNote()

Fetch Note from blockchain, decode R4-R7 registers, return NoteInfo with isExpired flag.

redeemNote()

Spend a Note, release ERG to receiver. Injects context variable 0 for predicate verification.

settleBatch()

Redeem multiple Notes in a single transaction with per-input context variables.

deployTracker()

Deploy an anti-double-spend Tracker box with empty spent set.

Full API reference: docs/api-reference.md

Basis: Off-Chain Credit, On-Chain Settlement

On-chain payments cost ERG and take seconds to confirm. For high-frequency agent commerce that's not always the right tradeoff. Basis is the off-chain credit layer documented in the repo:

Off-chain IOUs — payments happen between trusted parties, low fees, instant.
Credit creation — issuers can create unbacked notes within a trust relationship.
Optional reserves — when trust isn't enough, on-chain Reserves back the credit.
Global settlement — redeem against on-chain reserves whenever needed.

Read Basis docs

What's Next

Roadmap for the rest of 2026. PRs welcome — this is open source.

Production audit of ergo-agent-pay

Get the SDK reviewed by an external security firm before recommending it for mainnet credit issuance.

Standardize agent payment discovery

An EIP/RFC-style spec for how agents advertise payment requirements (think: well-known endpoint or 402 response header).

Reference dApp on testnet

Hosted demo agent that anyone can pay; full source code published; serves as the canonical 'this is what an agentic dApp looks like.'

More framework integrations

DSPy, LangGraph, Vercel AI SDK, LlamaIndex — wherever AI agents are being built, ergo-agent-pay should be one npm/pip command away.

Try It Now

Clone, install, run — under 5 minutes.

git clone https://github.com/bez111/ergo-agent-economy
cd ergo-agent-economy/examples/01-basic-payment
npm install
npm run start

The first example sends a testnet ERG payment between two agents and prints the transaction ID. From there, walk through 02 (Notes), 03 (acceptance predicates), and 04 (orchestrator) to see the full agent payment stack.

Clone the Repo Quickstart Guide Architecture

FAQ

What changed in Ergo's agent economy stack since the March 2026 launch?

The ergo-agent-pay SDK shipped v0.2 with the full Note lifecycle. Three packages now exist (TypeScript, Python, MCP server). Ten working examples ship with the repo, including LangChain, OpenAI, CrewAI, and AutoGen integrations.

Which AI agent frameworks now have native Ergo payment support?

LangChain, OpenAI function calling, CrewAI, and AutoGen all have working examples. Plus an MCP server so any MCP-compatible client (Claude Desktop, Cursor) can pay on Ergo testnet.

Is the agent payment stack production-ready?

Ergo protocol primitives are live on mainnet since 2019. The ergo-agent-pay SDK and ChainCash reference implementation are open-source prototypes — no security audits or production releases yet. Treat them as pre-production code, suitable for testnet development.

Where can I clone and run the examples?

github.com/bez111/ergo-agent-economy. Each example is self-contained with a README and runs against Ergo testnet — no mainnet ERG required.

What is the Basis layer?

Off-chain credit and IOU layer built on top of Ergo's on-chain primitives. Off-chain Notes circulate cheaply between trusted parties; on-chain Reserves act as collateral and global settlement when trust isn't enough.

Share this post

Help spread the word about Ergo's innovative blockchain technology

Quarterly updates on the Ergo agent economy stack: new packages, new framework adapters, what changed in the SDK.

Follow for daily updates