A
SpendGuard
Sign In
Get Started
Ecosystem

Agent Integrations

Plug-and-play financial capabilities for the world's leading AI frameworks. Give agents scoped payment capability through short-lived tokens and mandate checks.

Frameworks

Built for Every Stack

Use execution-token providers for LangChain, CrewAI, AutoGen, and function-calling agents.

LangChain

Most Popular

Add governed sandbox payment capability to LangChain agents with short-lived execution tokens.

langchain.py
import os
from spendguard import SpendGuardClient
from spendguard.integrations.langchain_tool import SpendGuardTool

management = SpendGuardClient(api_key=os.environ["SPENDGUARD_MANAGEMENT_KEY"])

def mint_execution_token():
    token = management.mint_execution_token(
        agent_id="agent_...",
        mandate_id="mandate_...",
    )
    return token["executionToken"]

payment_tool = SpendGuardTool(
    execution_token_provider=mint_execution_token,
    mandate_id="mandate_...",
)

agent = initialize_agent(
    tools=[payment_tool],
    llm=ChatOpenAI(model="gpt-4")
)

CrewAI

Growing Fast

Equip CrewAI agents with a governed payment tool that requests a fresh execution token per run.

crewai.py
import os
from crewai import Agent, Task
from spendguard import SpendGuardClient
from spendguard.integrations.crewai_tool import SpendGuardCrewTool

management = SpendGuardClient(api_key=os.environ["SPENDGUARD_MANAGEMENT_KEY"])

def mint_execution_token():
    token = management.mint_execution_token("agent_...", "mandate_...")
    return token["executionToken"]

pay_tool = SpendGuardCrewTool(
    execution_token_provider=mint_execution_token,
    mandate_id="mandate_...",
)

procurement_agent = Agent(
    role="Procurement Specialist",
    goal="Purchase required cloud resources",
    tools=[pay_tool],
    backstory="Expert at sourcing infrastructure"
)

task = Task(
    description="Purchase GPU instances for training",
    agent=procurement_agent
)

OpenAI Assistants

Beta

Expose a payment function that mints a short-lived token before calling the SpendGuard payment API.

openai assistants.py
import os
import openai
from spendguard import SpendGuardClient

management = SpendGuardClient(api_key=os.environ["SPENDGUARD_MANAGEMENT_KEY"])

def execute_payment(args):
    token = management.mint_execution_token("agent_...", "mandate_...")
    agent_spend = SpendGuardClient(api_key=token["executionToken"])
    return agent_spend.execute_payment(
        mandate_id="mandate_...",
        idempotency_key=args["idempotency_key"],
        **args,
    )

# Define as an OpenAI function
functions = [{
    "name": "execute_payment",
    "description": "Execute a payment via SpendGuard",
    "parameters": {
        "type": "object",
        "properties": {
            "amount": {"type": "number"},
            "merchant": {"type": "string"},
            "description": {"type": "string"}
        }
    }
}]

Autogen

New

Register SpendGuard with AutoGen using an execution-token provider instead of embedding raw API keys in tools.

autogen.py
import os
from autogen import AssistantAgent
from spendguard import SpendGuardClient
from spendguard.integrations.autogen_tool import register_spendguard

management = SpendGuardClient(api_key=os.environ["SPENDGUARD_MANAGEMENT_KEY"])

def mint_execution_token():
    token = management.mint_execution_token("agent_...", "mandate_...")
    return token["executionToken"]

finance_agent = AssistantAgent(
    name="FinanceBot",
    system_message="You handle payments...",
    llm_config={"model": "gpt-4"}
)

register_spendguard(
    finance_agent,
    execution_token_provider=mint_execution_token,
    mandate_id="mandate_...",
)
LangChain
CrewAI
OpenAI
Microsoft AutoGen
Semantic Kernel
Haystack

Request a Custom Integration

Working on a custom framework or need enterprise-specific features? Our team is here to help.