Skip to main content

The Payment Layer for AI Agents – Visa for the AI Age

Project description

SwarmPay -- The Payment Layer for AI Agents

PyPI Python 3.10+ License: MIT Tests Base Mainnet

AI agents autonomously pay each other via blockchain. Live on Base Mainnet with real USDC.

SwarmPay is the payment infrastructure SDK that lets AI agents discover services, negotiate prices, and transact autonomously using USDC on Base (Ethereum L2). Built-in on-chain escrow, reputation tracking, and spending controls.

Escrow Contract: 0x9Be790E8583f0425D213bF862148E8e531700106 (Base Mainnet, verified)

Install

pip install swarmpay

2-Minute Start

pip install swarmpay
swarmpay quickstart
swarmpay doctor
swarmpay demo

This creates a local demo-ready config first, so users can understand the flow before touching live blockchain credentials.

Quick Start

from swarmpay import SwarmPay

sp = SwarmPay(network="base-mainnet")
wallet = sp.create_wallet("my-agent")
result = wallet.pay("0xRecipient...", amount=0.50)  # Pays $0.50 USDC
print(result.explorer_url)  # basescan.org/tx/...

3 lines. No blockchain knowledge needed.

How It Works

Agent A (Customer)                    Agent B (Provider)
    |                                     |
    |-- 1. Discover -- SwarmPay Registry -|
    |-- 2. Pay $0.50 -- Base Blockchain --|
    |       $0.495 -> Agent B (USDC)      |
    |       $0.005 -> SwarmPay Fee (1%)   |
    |-- 3. Receive -- Service Result -----|
  1. Agent B registers a service (e.g., "text-analysis @ $0.50")
  2. Agent A discovers it via the Service Registry
  3. Agent A pays automatically -- USDC splits between provider (99%) and fee (1%)
  4. Agent B delivers the service

Setup

# Fastest path
swarmpay quickstart
swarmpay doctor
swarmpay demo

# When you're ready for live payments
swarmpay setup

Features

Feature Description
3 Lines of Code Import, initialize, pay. No blockchain knowledge needed.
Base Mainnet Live on Base with real USDC. Testnet also supported.
Escrow Protection Funds held until service delivery is confirmed.
Reputation System Track agent reliability with score-based provider selection.
Spending Controls Per-transaction, daily, and monthly limits with service filters.
Non-Custodial We never touch funds. Coinbase TEE-secured wallets.
Service Registry Agents register and discover services automatically.
Volume Discounts Fees decrease from 1% to 0.3% based on monthly volume.
Framework Agnostic Works with LangChain, CrewAI, AutoGen, OpenAI, Claude MCP.

On-Chain Escrow

Funds are held in a verified smart contract until service delivery is confirmed. No trust required.

Buyer                     Contract                    Provider
  |-- approve USDC -------->|                            |
  |-- fundEscrow() -------->| holds funds                |
  |                         |<-- confirmDelivery() ------|
  |-- releaseEscrow() ----->| pays provider + fee        |

SDK Usage

from swarmpay import SwarmPay, EscrowConfig

sp = SwarmPay(
    network="base-mainnet",
    escrow_config=EscrowConfig(default_timeout_seconds=3600),
)

# Buyer purchases with escrow protection
result = sp.buy_service_escrow(buyer_wallet, "translation")

# Provider confirms delivery
sp.confirm_delivery(result["escrow"].escrow_id, provider_address)

# Buyer releases funds (or disputes)
sp.release_escrow(result["escrow"].escrow_id)

CLI

swarmpay escrow list                              # List active escrows
swarmpay escrow fund <provider> <amount>          # Create escrow
swarmpay escrow confirm <escrow_id>               # Provider confirms delivery
swarmpay escrow release <escrow_id>               # Buyer releases funds
swarmpay escrow dispute <escrow_id> <reason>      # Open dispute
swarmpay escrow refund <escrow_id>                # Refund (after timeout/dispute)

Reputation System

from swarmpay import SwarmPay, ReputationConfig

sp = SwarmPay(reputation_config=ReputationConfig())

# Automatically tracks success/failure/disputes
# Score: 0-5 (new agents start at 3.0)

# find_cheapest() now considers reputation + price
best = sp.find_cheapest("translation")

# Check any agent's reputation
rep = sp.get_reputation("0xAgentWallet...")
print(f"Score: {rep.score}, Deliveries: {rep.total_deliveries}")

Spending Controls

from swarmpay import SwarmPay, PermissionEngine, SpendingLimits, AgentPermissions

perms = PermissionEngine()
perms.set_permissions("0xAgentWallet...", AgentPermissions(
    spending_limits=SpendingLimits(
        per_transaction_usd=100.0,
        daily_usd=500.0,
        monthly_usd=5000.0,
    ),
    allowed_service_types=["translation", "analysis"],
))

sp = SwarmPay(permission_engine=perms)

Integrations

REST API

export SWARMPAY_API_KEY="your-api-key"
python -m swarmpay.api_server  # Starts on port 8420

curl http://localhost:8420/pay -X POST \
  -H "Authorization: Bearer $SWARMPAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"agent_id": "my-agent", "recipient": "0x...", "amount": 0.50}'

MCP Server (Claude)

python -m swarmpay.mcp_server
# Tools: swarmpay_create_wallet, swarmpay_pay, swarmpay_escrow_fund,
#         swarmpay_escrow_release, swarmpay_escrow_dispute, etc.

OpenAI Function Calling

from swarmpay.openai_tools import SWARMPAY_TOOLS, handle_swarmpay_call

tools = SWARMPAY_TOOLS
response = handle_swarmpay_call(
    "swarmpay_buy_service",
    {"service_type": "translation", "max_price_usd": 0.50},
    swarmpay_client=sp,
    wallet=wallet,
)

CLI

swarmpay quickstart   # Create local demo config
swarmpay doctor       # Check demo/live readiness
swarmpay setup        # Interactive setup
swarmpay status       # Check wallet & network
swarmpay demo         # Interactive demo
swarmpay serve        # Start API server
swarmpay escrow list  # Manage escrows

Architecture

swarmpay/
├── __init__.py         # Package exports (v0.4.2)
├── client.py           # Main SwarmPay class
├── wallet.py           # AgentWallet -- USDC & ETH transfers
├── constants.py        # USDC contracts, escrow addresses, network config
├── fee_engine.py       # Volume-based fee tiers
├── registry.py         # Service discovery
├── escrow.py           # Off-chain escrow engine
├── onchain_escrow.py   # On-chain escrow (Base Mainnet contract)
├── reputation.py       # Agent reputation scoring
├── permissions.py      # Spending limits & access control
├── webhooks.py         # Webhook event notifications
├── dashboard.py        # Dashboard feed & summaries
├── api_server.py       # REST API (port 8420, auth + rate limiting)
├── mcp_server.py       # Claude MCP integration
├── cli.py              # CLI tool
└── openai_tools.py     # OpenAI function calling

contracts/
└── SwarmpayEscrow.sol  # Solidity escrow contract (verified on Basescan)

tests/                  # 168 unit + integration tests

Fee Structure

Monthly Volume Fee
$0 -- $10K 1.0%
$10K -- $100K 0.8%
$100K -- $1M 0.5%
$1M+ 0.3%

Networks

Network ID Status
Base Mainnet base-mainnet Live
Base Sepolia base-sepolia Testnet

Tech Stack

  • Blockchain: Base (Ethereum L2 by Coinbase)
  • Currency: USDC (stablecoin, 6 decimals)
  • Wallet Infrastructure: Coinbase CDP SDK (AgentKit)
  • Security: Non-custodial, TEE-secured wallets, Bearer auth, rate limiting
  • Smart Contract: Solidity 0.8.24, verified on Basescan
  • Protocol: ERC-20 token transfers + on-chain escrow

Running Tests

pip install swarmpay[dev]
python -m pytest tests/ -v  # 168 tests

License

MIT -- see LICENSE

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

swarmpay-0.4.3.tar.gz (54.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

swarmpay-0.4.3-py3-none-any.whl (47.2 kB view details)

Uploaded Python 3

File details

Details for the file swarmpay-0.4.3.tar.gz.

File metadata

  • Download URL: swarmpay-0.4.3.tar.gz
  • Upload date:
  • Size: 54.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.13

File hashes

Hashes for swarmpay-0.4.3.tar.gz
Algorithm Hash digest
SHA256 e4fb417ceaca5cd657df3f44e03a861a556a73e90d7794fe1d07eb4f3a736ed4
MD5 78cbde902f3e2cfc0a322c9d50e5da42
BLAKE2b-256 7ce040842722a36337b82390bae6525ba2593f2c4d97b287ff012fb08b8380e3

See more details on using hashes here.

File details

Details for the file swarmpay-0.4.3-py3-none-any.whl.

File metadata

  • Download URL: swarmpay-0.4.3-py3-none-any.whl
  • Upload date:
  • Size: 47.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.13

File hashes

Hashes for swarmpay-0.4.3-py3-none-any.whl
Algorithm Hash digest
SHA256 5592e37e7359b92a69e9a21bc5d3f2d4618db78a3b40f169924922b484788115
MD5 2c28fbb0e5755ce69cdd5637c1c0cbaf
BLAKE2b-256 540025d57c9e66e5d4a9f1a2eaa826ac1492fb3a6940fc1c0dda593cc3132b2c

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page