Skip to main content

LangChain integration for PQSafe AgentPay — post-quantum safe payments for AI agents

Project description

langchain-pqsafe

PyPI License: MIT

Drop post-quantum signed payments into any LangChain agent in one line — FIPS 204 (ML-DSA-65) enforced.

Part of the PQSafe AgentPay ecosystem. Built on pqsafe-agent-pay.


What it does

langchain-pqsafe provides PQSafePaymentTool, a LangChain BaseTool that gives your agent the ability to make payments authorized by a signed SpendEnvelope — a post-quantum (ML-DSA-65, NIST FIPS 204) token issued by the human wallet owner. Add the tool to your agent's tool list and every pqsafe_pay tool call is automatically signature-verified and policy-enforced before any payment is dispatched.

No long-lived API keys in your agent code. No credentials to rotate. The envelope constrains exactly what the agent can spend, to whom, via which rail, and for how long — and it cannot be forged or exceeded.


Install

pip install langchain-pqsafe

Quickstart

from langchain_pqsafe import PQSafePaymentTool
from langchain.agents import create_react_agent
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(model="gpt-4o")
agent = create_react_agent(llm, tools=[PQSafePaymentTool()])

That's it. The agent can now call pqsafe_pay whenever it needs to make a payment authorized by a SpendEnvelope.


Full example

import json
from langchain_pqsafe import PQSafePaymentTool
from langchain.agents import AgentExecutor, create_react_agent
from langchain_openai import ChatOpenAI
from langchain_core.prompts import PromptTemplate

# 1. Build the tool
payment_tool = PQSafePaymentTool()

# 2. Your signed envelope (issued by the wallet owner, stored securely)
signed_envelope = {
    "envelopeJson": '{"version":1,"issuer":"pq1...","agent":"research-agent-v1",...}',
    "signature": "deadbeef...",
    "dsaPublicKey": "cafebabe...",
}
envelope_json = json.dumps(signed_envelope)

# 3. Create the agent — the envelope is injected into the prompt context
llm = ChatOpenAI(model="gpt-4o", temperature=0)
prompt = PromptTemplate.from_template(
    "You are a payment agent. Use pqsafe_pay to pay invoices.\n"
    "Envelope: {envelope_json}\n\n{agent_scratchpad}"
)
agent = create_react_agent(llm, tools=[payment_tool], prompt=prompt)
executor = AgentExecutor(agent=agent, tools=[payment_tool], verbose=True)

# 4. Run — the agent will call pqsafe_pay with the envelope automatically
result = executor.invoke({
    "envelope_json": envelope_json,
    "input": "Renew Perplexity Pro at perplexity.ai — $20 USD for research-agent-v1",
})
print(result["output"])
# Payment successful. txId=airwallex-tx-abc123 status=success rail=airwallex

How it works

  1. A human wallet owner issues a signed SpendEnvelope using pqsafe-agent-pay (or the PQSafe wallet). The envelope encodes: agent ID, max amount, allowed recipients, currency, and validity window — all bound by an ML-DSA-65 post-quantum signature.
  2. The envelope is passed to the LangChain agent as prompt context (or tool input). It travels with the agent's context, not stored in the tool.
  3. When the LLM decides to pay, it calls the pqsafe_pay tool with envelope_json, recipient, amount, and optional memo.
  4. PQSafePaymentTool verifies the post-quantum signature server-side, enforces all policy constraints, and routes the payment to the cheapest available rail (Airwallex, Wise, Stripe, USDC/Base, or x402).
  5. The tool returns a string: Payment successful. txId=<id> status=<status> rail=<rail> — which the LLM can include in its final answer.

What you get

  • One-line integrationtools=[PQSafePaymentTool()] is all it takes
  • FIPS 204 ML-DSA-65 enforcement — every payment is quantum-resistant by default
  • Policy guardrails — amount ceiling, recipient allowlist, and time window enforced before dispatch; the LLM cannot overspend
  • Multi-rail routing — Airwallex (live sandbox), Wise, Stripe, USDC on Base, x402
  • No credentials in agent code — only the short-lived signed envelope is required at runtime
  • Compatible with any LangChain agent — ReAct, OpenAI Functions, Structured Chat, and custom agent types

Tool schema

Parameter Type Required Description
envelope_json str Yes SignedEnvelope serialized as JSON
recipient str Yes Recipient address (IBAN, crypto addr, domain, etc.)
amount float Yes Amount in the envelope's currency (> 0)
memo str No Human-readable payment reference

Returns a string: Payment successful. txId=<id> status=<status> rail=<rail>


Part of PQSafe AgentPay


Links


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

langchain_pqsafe-0.1.1.tar.gz (7.2 kB view details)

Uploaded Source

Built Distribution

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

langchain_pqsafe-0.1.1-py3-none-any.whl (6.7 kB view details)

Uploaded Python 3

File details

Details for the file langchain_pqsafe-0.1.1.tar.gz.

File metadata

  • Download URL: langchain_pqsafe-0.1.1.tar.gz
  • Upload date:
  • Size: 7.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.9.6

File hashes

Hashes for langchain_pqsafe-0.1.1.tar.gz
Algorithm Hash digest
SHA256 0ce5d3e4faa8dbf8092ffc53dc26657e9a87b1fd98ae484d483e263305fb9859
MD5 6474c9d469b317a877aadf19b775016d
BLAKE2b-256 791b0801281adca164328547624cb40680a34447b0ac2239394c22e602c4f139

See more details on using hashes here.

File details

Details for the file langchain_pqsafe-0.1.1-py3-none-any.whl.

File metadata

File hashes

Hashes for langchain_pqsafe-0.1.1-py3-none-any.whl
Algorithm Hash digest
SHA256 27bf75a531c6f1bd69c5fe544e9a4b675a854d0027ea3663c0f6d4a60ee0ed31
MD5 38d4f3f99368900fdd0f9160258d4099
BLAKE2b-256 c188475b3cab305d4d4635ec59a210cb8541dd8939830fde50ba1766dc3f47f3

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