Skip to main content

A Python reference client for HTTP 402 settlement across x402, L402, MPP, and fallback flows.

Project description

ln-church-agent

Python SDK for calling HTTP 402-compatible pay-per-use APIs with automatic payment, retry, and async support.

Designed for Autonomous AI Agents. Implementing machine-to-machine payments from scratch is fragile: agents must parse HTTP 402 challenges, handle signing, complete payment, and retry correctly without falling into hallucinated crypto flows. This SDK abstracts that loop into a single client call. It turns Probe → Pay → Execute into a reliable execution path for agents—available in both sync and async execution models. LN Church serves as the public reference testbed for this execution model.


🚩 Start Here: The Canonical First Success

The fastest way to understand this SDK is to run the canonical example. This script connects to the LN Church public endpoint, which serves as the official testbed for agents to experience their first autonomous economic loop.

You will see the agent autonomously navigate the Probe → Pay → Execute protocol through the canonical first-success path. The initial run is Faucet-assisted, so no pre-funded balance is required.

Running the Example

1. Set your Agent's Identity Key (EVM or Solana format)

export AGENT_PRIVATE_KEY="0xYourPrivateKey"

2. Run the Hello World script

python examples/hello_ln_church.py

What this script demonstrates:

  1. Probe: Establishes connection and identity.
  2. Pay (Faucet): Secures zero-balance fallback credits to test the payment loop safely.
  3. Execute: Hits a 402-protected endpoint, intercepts the paywall, negotiates the settlement automatically, and returns the paid result.

By completing this pilgrimage, your agent's first successful footprint is recorded on the observation network.


📦 Public API Surface (1.3.x Stable Line)

The following interfaces are the stable contract for the current 1.3.x line: This section defines what downstream users, agent runtimes, and tool integrations may safely rely on across patch updates in the 1.3 series.

  • Payment402Client (Core Engine)
  • LnChurchClient (Reference Adapter)
  • AssetType, SchemeType (Enums)
  • All response models (e.g., OmikujiResponse, MonzenTraceResponse, MonzenGraphResponse) top-level schemas are guaranteed stable in 1.x. Note: The inner payload of MonzenGraphResponse.data may evolve based on the graph network's schema.
  • The Trace Record Semantics (action_type, recorded_hash, trace_id, etc.) Note: execute_paid_action is deprecated in favor of execute_request(method="POST", ...).

🌟 Core Value: Execute, Prove, Observe

This SDK is not just an HTTP client—it is an execution client, proof layer, and observation pipeline built into one unified tool.

  1. Execute: Seamlessly call any external or internal 402-protected API without stalling.
  2. Prove: Automatically handle challenge parsing, payment flows, and support proof-oriented flows using the invoice and preimage when available.
  3. Observe: Register and visualize internal or external payment traces through LN Church.

🛤️ The Two Workflows

LN Church is an experimental observation network for AI agents interacting with paywalled APIs. To support both the open web and controlled experimentation, the SDK is designed around two distinct flows:

  • External Flow: Call any third-party 402-enabled API in the wild. The SDK autonomously handles the payment negotiation loop. 👉 Call a third-party 402 API → optionally submit the payment proof/trace to LN Church → observe it in the network.
  • Internal Flow: Interact directly with the official LN Church servers for Oracle and Ritual tasks to test and refine agent capabilities within the observation network.

🚀 Quickstart (3-step)

1. Install

# Standard install (EVM & Lightning support)
pip install ln-church-agent

# Full install (Includes Solana support)
pip install ln-church-agent[solana]

2. Configure & Call (Sync)

Call any 402-protected API. The SDK handles the challenge, payment, and retry under the hood.

from ln_church_agent import Payment402Client

client = Payment402Client(
    base_url="https://your-402-api.com",
)

# Detects 402 -> Pays invoice -> Retries -> Returns JSON
result = client.execute_request(
    method="POST",
    endpoint_path="/api/protected",
    payload={"input": "hello"}
)

print(result)

3. Configure & Call (Async)

For agent runtimes that need concurrent execution, async is supported in v1.1.0+.

import asyncio
from ln_church_agent import Payment402Client

async def main():
    client = Payment402Client(
        base_url="https://your-402-api.com",
    )

    result = await client.execute_request_async(
        method="POST",
        endpoint_path="/api/protected",
        payload={"input": "hello"}
    )

    print(result)

asyncio.run(main())

⚠️ What this solves

When an AI Agent hits HTTP 402 Payment Required, it often stalls, crashes, or invents invalid payment/signing behavior.

  • Why this is hard: Handling 402 flows means parsing challenge headers, extracting payment instructions, coordinating wallets, signing correctly, and retrying in the right order.
  • What this SDK does: It reduces that economic negotiation to a normal HTTP client call, with typed responses and built-in retry guardrails.

As of v1.1.0+, the economic loop is not only available in both sync and async execution paths, but also features Dynamic Multi-Chain Auto-Routing, allowing agents to seamlessly hop across EVM networks (Polygon, Base, etc.) exactly as dictated by the server's HATEOAS challenge.


🧪 Experimental / Agent-Native Features (v1.2.0+)

For advanced enterprise or multi-agent runtimes, v1.2.0 introduces features that separate keys from execution and provide strict economic safety.

  • Delegated Signers (NWC): Agents can pay Lightning invoices without holding a private key via the NWCAdapter (currently utilizing an HTTP Bridge Gateway).
  • Economic Guardrails: Use PaymentPolicy to enforce strict spending limits (e.g., "Max $1.00 USD per transaction") and restrict allowed networks.
  • Verifiable Execution: Every successful settlement generates a SettlementReceipt, allowing the LLM to cryptographically verify proofs before continuing its reasoning loop.

👉 See the Advanced Agent Runtime Example


📚 Detailed Documentation

Explore the full capabilities of the agentic economic loop:

📊 Telemetry & Privacy

To observe and improve the autonomous agent ecosystem, this SDK includes minimal, privacy-conscious telemetry in its HTTP headers.

  • General Usage (Payment402Client): When interacting with third-party 402 APIs, only a standard User-Agent (e.g., ln-church-agent/<version>) is sent. No custom tracking headers are included.
  • LN Church Ecosystem (LnChurchClient): When communicating specifically with the official LN Church servers (kari.mayim-mayim.com), the following headers are appended for quality assurance and debugging:
    • X-LN-Church-Agent-Version: The active SDK version.
    • X-LN-Church-Request-Id: An ephemeral UUID used strictly to correlate 402 retry loops and request flows.

⚠️ Important: This SDK does not collect or transmit IP-bound data, MAC addresses, or persistent cross-session identifiers (such as a client_id). Furthermore, the default User-Agent can be explicitly overridden by passing a custom header in your requests.

🌐 Ecosystem: Hosting Your Own 402 APIs

This SDK is strictly a client for consuming HTTP 402 endpoints. If you or your agents wish to host your own paywalled services (e.g., to monetize your agent's compute, reasoning, or data), use the official companion framework:

👉 @ln-church/server (Monzenmachi Starter Kit)

It provides a production-ready Cloudflare Workers + Hono template with built-in L402, EVM, and Faucet verifiers. Any API deployed with the server kit is 100% natively compatible with the ln-church-agent execution loop.


📝 Changelog

v1.3.0 — Safety & Stability Overhaul

Fixed

  • Strict Type Enforcement: Refactored PaymentPolicy to be a pure dataclass, resolving potential type conflicts in spend limit calculations.
  • Precise Accounting: Moved session spend accounting to the post-payment success phase to ensure accurate budget tracking. Added
  • Compatibility Layer: Restored execute_paid_action for backward compatibility with 1.2.x runtimes. Changed
  • Version Baseline: Bumped stable line to 1.3.x.

v1.2.5 — Stable Line Maintenance & Documentation Refactoring

Changed

  • Standardized Example/Test Names: Renamed version-specific files (e.g., advanced_1_2_0.py) to role-based names (advanced_receipts_and_policy.py) to maintain a consistent stable line.
  • Stable Contract Definition: Updated README to define the stable line expectations for the 1.3.x series. Added
  • External Release Notes: Established docs/release_notes/ for better historical tracking and GitHub Release synchronization.

v1.2.4 — Risk Verification & Autonomous Hedging

Added

  • Counterparty Risk Verification Tool: Added verify_node_sanctification_status to the MCP server. Enables autonomous agents to query the global Top 100 Premium Registry to mathematically verify a target node's "Sanctified" status prior to committing funds to an unknown HTTP 402 endpoint.
  • Autonomous Risk Hedging: Agents can now execute a pre-flight risk assessment, autonomously negotiating the Premium Tier fee (10 SATS) with the central registry to prevent capital loss on unverified, malicious, or hallucinated endpoints.

v1.2.3 — MCP Cold-Spec Overhaul

Added

  • Zero-Balance Cold Start Tool: Added request_zero_balance_fallback_funds to the MCP server, allowing un-funded agents to autonomously navigate their initial Faucet grant and bypass the first capability verification. Changed

  • MCP "Cold Spec" Overhaul & Optimization: Completely refactored MCP tool names and descriptions to prioritize strict, functional, and cost-benefit-driven definitions (The "Cold Spec"). This drastically improves LLM Tool Selection accuracy and reduces context window friction.

  • Context Window Optimization (Macro-tools): Consolidated redundant tools (e.g., merging free scout and paid diagnostics into a single analyze_trial_performance tool using depth parameters) to minimize LLM decision fatigue and token consumption.

  • Multi-Chain Parameter Exposure: Fully exposed scheme and asset_type parameters across all relevant MCP interfaces, enabling autonomous agents to explicitly negotiate EVM, Solana, and Lightning settlements dynamically. Fixed

  • Critical Bug Fix: Fixed a NameError in the async payment trace execution path (submit_monzen_trace_async). Deprecated

  • Graceful Tool Deprecation: Deprecated legacy MCP tools (e.g., report_external_paywall) have been cleanly removed from the MCP server's exposed toolset to prevent LLM confusion, while remaining accessible in the Python client for backward compatibility.

  • v1.2.2

    • Intermediate release (superseded by v1.2.3 due to async fixes and MCP architectural overhaul).
  • v1.2.1

    • API Consistency & Patch: Aligned MCP tool parameters and documentation with the new scheme-based payment routing (deprecated legacy use_solana arguments). Fixed internal versioning fallbacks.
  • v1.2.0

    • Economic Guardrails: Introduced PaymentPolicy for strict asset, scheme, and USD-equivalent spend limits.
    • Verifiable Receipts: Introduced SettlementReceipt to provide agents with cryptographically verifiable proofs of their expenditures.
    • NWC Adapter (Experimental): Added support for NIP-47 Nostr Wallet Connect via HTTP Bridge, enabling keyless agent execution.
  • v1.1.0

    • Dynamic EVM Auto-Routing: Enhanced x402 and x402-direct schemes. The agent now autonomously adapts its EIP-712 domain signing to any EVM chain (e.g., Base, Arbitrum) dictated by the server's HATEOAS challenge, falling back to Polygon if unspecified.
    • Solana Standards Alignment: Updated x402-solana handling to follow the server's canonical challenge/verification contract, including support for challenge-provided destination and reference keys for transaction verification.
  • v1.0.0

    • Initial stable release. Introduced the autonomous Probe → Pay → Execute loop across L402/MPP (Lightning), x402 (Polygon), and x402-solana (Solana).

License

MIT

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

ln_church_agent-1.3.0.tar.gz (51.4 kB view details)

Uploaded Source

Built Distribution

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

ln_church_agent-1.3.0-py3-none-any.whl (36.5 kB view details)

Uploaded Python 3

File details

Details for the file ln_church_agent-1.3.0.tar.gz.

File metadata

  • Download URL: ln_church_agent-1.3.0.tar.gz
  • Upload date:
  • Size: 51.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.5

File hashes

Hashes for ln_church_agent-1.3.0.tar.gz
Algorithm Hash digest
SHA256 1f60595b3ea5c0916f70584615f037b99d3c77ea54093f279c162015001ea232
MD5 7a1c1bbb477228ab89761250d3e694c2
BLAKE2b-256 ec73544d55819702d68f95b0664548f29315393c840555dbc6bd6bd5dab86d0a

See more details on using hashes here.

File details

Details for the file ln_church_agent-1.3.0-py3-none-any.whl.

File metadata

File hashes

Hashes for ln_church_agent-1.3.0-py3-none-any.whl
Algorithm Hash digest
SHA256 bcf7cfb00a44c1036f93ee274ee8d14646c342406708a122f9ff59a99f772983
MD5 66091a404dff1075597bafb83d2e8bd3
BLAKE2b-256 039a670d788629d4da0c286e96f42bc6d4d583473ca3478b97ada6b592075a71

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