LangChain and CrewAI Agents Can Now Execute Compliant Payments

LangChain agents plan, research, and decide. Until now, every workflow that required spending stopped at the payment step. Either a human completed the transaction, or the agent operated with unconstrained access to a live account.

Today that changes. langchain-oris and crewai-oris are live on PyPI.

What the packages do

Both packages wrap the Oris SDK as native tool sets for their respective frameworks.

langchain-oris exposes seventeen LangChain tools across five categories. Payments: oris_pay, oris_check_balance, oris_get_spending, oris_approve_pending. Identity and KYA: oris_get_tier_info, oris_get_did, oris_get_eas_attestation, oris_generate_attestation, oris_promotion_status. Marketplace: oris_find_service, oris_place_order, oris_complete_order. Fiat bridge: oris_fiat_onramp, oris_fiat_offramp, oris_exchange_rate. Cross-chain: oris_cross_chain_quote. BYOK: oris_configure_provider for Turnkey, Fireblocks, Pimlico, Stripe, Circle, and Solana RPC. The OrisToolkit bundles all seventeen for one-line setup, and python -m oris_langchain setup runs an interactive wizard that registers a fresh agent and writes a .env file. Every tool supports both sync (_run) and async (_arun) execution patterns, making it compatible with LangGraph node graphs that mix sync and async steps.

crewai-oris exposes the same seventeen capabilities as CrewAI BaseTool subclasses with human-readable names like Make Compliant Payment, Check Agent Balance, Get Agent DID, Configure Provider Keys (BYOK), and Cross-chain Quote. The OrisToolset bundles all seventeen at once, and python -m oris_crewai setup registers a fresh agent in under a minute. Credentials read from environment variables by default, which fits the CrewAI deployment pattern cleanly.

What happens when an agent calls oris_pay

The call triggers the full Oris pipeline before anything touches the blockchain:

If any stage fails or is unreachable, the transaction is blocked. The agent receives a structured error string it can reason about and report.

Quick start: LangChain

from oris_langchain import OrisToolkit
from langchain_openai import ChatOpenAI
from langgraph.prebuilt import create_react_agent

toolkit = OrisToolkit(
    api_key="oris_sk_live_...",
    private_key="your_private_key",
    agent_id="your_agent_uuid",
)

llm = ChatOpenAI(model="gpt-4o")
agent = create_react_agent(llm, toolkit.get_tools())

result = agent.invoke({
    "messages": [("user", "Pay 10 USDC to 0xABC... for the API subscription.")]
})

Quick start: CrewAI

from crewai import Agent, Task, Crew
from oris_crewai import OrisPaymentTool, OrisCheckBalanceTool

finance_agent = Agent(
    role="Finance Agent",
    goal="Execute compliant payments on behalf of the crew",
    backstory="You handle all financial transactions with KYA compliance.",
    tools=[OrisPaymentTool(), OrisCheckBalanceTool()],
)

task = Task(
    description="Pay 25 USDC to 0xABC... for the API subscription renewal.",
    expected_output="Payment confirmation with transaction hash",
    agent=finance_agent,
)

Crew(agents=[finance_agent], tasks=[task]).kickoff()

Installation

$ pip install langchain-oris
$ pip install crewai-oris

Both packages require Python 3.10+ and an Oris API key. Get started at useoris.xyz.

Full documentation: docs.useoris.xyz

PyPI: langchain-oris | crewai-oris