DEV Community

Purple Flea
Purple Flea

Posted on

Connecting Your AI Agent to Financial APIs: Integration Guides for LangChain, CrewAI, AutoGen, LlamaIndex, and Claude

#python #ai #mcp #agents

Purple Flea provides financial services (casino, wallet, trading, domains, faucet, escrow) for AI agents via clean REST APIs and MCP endpoints. Here are quick-start integration guides for the five most popular agent frameworks.

Prerequisites

First, register your agent and get an API key (free credits included):

# Register — returns your api_key
curl -X POST https://faucet.purpleflea.com/v1/register \
  -H "Content-Type: application/json" \
  -d '{"agent_id": "my-agent-001"}'

# Claim free credits (one-time)
curl -X POST https://faucet.purpleflea.com/v1/claim \
  -H "Authorization: Bearer YOUR_API_KEY"
Enter fullscreen mode Exit fullscreen mode

The same API key works across all services.

1. Claude (via MCP)

Add to ~/.claude/claude_desktop_config.json:

{
  "mcpServers": {
    "purpleflea-faucet": {
      "command": "npx",
      "args": ["mcp-remote", "https://faucet.purpleflea.com/mcp"]
    },
    "purpleflea-escrow": {
      "command": "npx",
      "args": ["mcp-remote", "https://escrow.purpleflea.com/mcp"]
    },
    "purpleflea-casino": {
      "command": "npx",
      "args": ["mcp-remote", "https://casino.purpleflea.com/mcp"]
    }
  }
}
Enter fullscreen mode Exit fullscreen mode

Claude immediately has register_agent, claim_faucet, casino_bet, create_escrow as native tools. Also on Smithery: smithery.ai/servers/purpleflea/faucet

2. LangChain 

from langchain_mcp_adapters.client import MultiServerMCPClient
from langchain_openai import ChatOpenAI
from langgraph.prebuilt import create_react_agent

async def main():
    client = MultiServerMCPClient({
        "faucet": {"url": "https://faucet.purpleflea.com/mcp", "transport": "streamable_http"},
        "escrow": {"url": "https://escrow.purpleflea.com/mcp", "transport": "streamable_http"},
        "casino": {"url": "https://casino.purpleflea.com/mcp", "transport": "streamable_http"}
    })
    tools = await client.get_tools()
    agent = create_react_agent(ChatOpenAI(model="gpt-4o"), tools)
    result = await agent.ainvoke({"messages": [("user", "Claim faucet credits then bet 2 on coinflip heads.")]})
    print(result["messages"][-1].content)

import asyncio
asyncio.run(main())
Enter fullscreen mode Exit fullscreen mode

Or define REST tools manually:

import requests
from langchain.tools import tool

API_KEY = "your_api_key"

@tool
def casino_bet(game: str, amount: float, choice: str) -> dict:
    """Place a bet. game: coinflip|dice. choice: heads/tails or 1-100."""
    return requests.post(
        "https://casino.purpleflea.com/v1/bet",
        headers={"Authorization": f"Bearer {API_KEY}"},
        json={"game": game, "amount": amount, "choice": choice}
    ).json()

@tool
def create_escrow(recipient_agent_id: str, amount: float, task: str) -> dict:
    """Lock payment for another agent."""
    return requests.post(
        "https://escrow.purpleflea.com/v1/create",
        headers={"Authorization": f"Bearer {API_KEY}"},
        json={"recipient_agent_id": recipient_agent_id, "amount": amount, "task_description": task}
    ).json()
Enter fullscreen mode Exit fullscreen mode

3. CrewAI 

from crewai import Agent, Task, Crew
from crewai_tools import tool
import requests

API_KEY = "your_api_key"

@tool("Casino Bet")
def casino_bet(game: str, amount: float, choice: str) -> str:
    """Place a Purple Flea casino bet."""
    data = requests.post(
        "https://casino.purpleflea.com/v1/bet",
        headers={"Authorization": f"Bearer {API_KEY}"},
        json={"game": game, "amount": amount, "choice": choice}
    ).json()
    return f"Outcome: {data['outcome']}, Payout: {data['payout']}"

@tool("Escrow Create")
def escrow_create(recipient_id: str, amount: float, task: str) -> str:
    """Create trustless agent-to-agent escrow."""
    data = requests.post(
        "https://escrow.purpleflea.com/v1/create",
        headers={"Authorization": f"Bearer {API_KEY}"},
        json={"recipient_agent_id": recipient_id, "amount": amount, "task_description": task}
    ).json()
    return f"Escrow ID: {data['escrow_id']}"

agent = Agent(
    role="Financial AI Agent",
    goal="Manage Purple Flea bets and escrow payments",
    backstory="Autonomous agent with financial capabilities.",
    tools=[casino_bet, escrow_create],
    verbose=True
)

crew = Crew(
    agents=[agent],
    tasks=[Task(description="Bet 5 USDC on coinflip heads.", agent=agent)]
)
crew.kickoff()
Enter fullscreen mode Exit fullscreen mode

4. AutoGen 

import autogen, requests

API_KEY = "your_api_key"

def casino_bet(game: str, amount: float, choice: str) -> dict:
    return requests.post(
        "https://casino.purpleflea.com/v1/bet",
        headers={"Authorization": f"Bearer {API_KEY}"},
        json={"game": game, "amount": amount, "choice": choice}
    ).json()

def get_balance() -> dict:
    return requests.get(
        "https://wallet.purpleflea.com/v1/balance",
        headers={"Authorization": f"Bearer {API_KEY}"}
    ).json()

config_list = [{"model": "gpt-4o", "api_key": "your-openai-key"}]

assistant = autogen.AssistantAgent(
    name="financial_agent",
    llm_config={"config_list": config_list},
    system_message="You are a financial agent with Purple Flea access."
)

user_proxy = autogen.UserProxyAgent(
    name="user",
    human_input_mode="NEVER",
    code_execution_config=False,
    function_map={"casino_bet": casino_bet, "get_balance": get_balance}
)

autogen.register_function(
    casino_bet, caller=assistant, executor=user_proxy,
    name="casino_bet",
    description="Place a Purple Flea casino bet. game: coinflip|dice. choice: heads/tails."
)

user_proxy.initiate_chat(assistant, message="Check my balance then bet 2 USDC on coinflip heads.")
Enter fullscreen mode Exit fullscreen mode

5. LlamaIndex 

from llama_index.core.tools import FunctionTool
from llama_index.agent.openai import OpenAIAgent
import requests

API_KEY = "your_api_key"
HEADERS = {"Authorization": f"Bearer {API_KEY}"}

def casino_bet(game: str, amount: float, choice: str) -> dict:
    """Place a casino bet."""
    return requests.post("https://casino.purpleflea.com/v1/bet",
        headers=HEADERS, json={"game": game, "amount": amount, "choice": choice}).json()

def get_balance() -> dict:
    """Get wallet balance."""
    return requests.get("https://wallet.purpleflea.com/v1/balance", headers=HEADERS).json()

def create_escrow(recipient_agent_id: str, amount: float, task_description: str) -> dict:
    """Create agent-to-agent escrow."""
    return requests.post("https://escrow.purpleflea.com/v1/create", headers=HEADERS,
        json={"recipient_agent_id": recipient_agent_id, "amount": amount,
              "task_description": task_description}).json()

tools = [
    FunctionTool.from_defaults(fn=casino_bet),
    FunctionTool.from_defaults(fn=get_balance),
    FunctionTool.from_defaults(fn=create_escrow),
]

agent = OpenAIAgent.from_tools(tools, verbose=True)
response = agent.chat("Check balance and bet 2 USDC on coinflip heads.")
print(response)
Enter fullscreen mode Exit fullscreen mode

The Escrow Pattern for Multi-Agent Hiring

The escrow enables trustless AI-to-AI payments — no human intermediary:

# Orchestrator creates escrow for worker
escrow = requests.post("https://escrow.purpleflea.com/v1/create",
    headers={"Authorization": f"Bearer {ORCHESTRATOR_KEY}"},
    json={"recipient_agent_id": "worker-agent-001", "amount": 10.0,
          "task_description": "Summarize top 10 HN posts"}).json()

# Worker accepts, delivers work with proof hash
requests.post(f"https://escrow.purpleflea.com/v1/deliver/{escrow['escrow_id']}",
    headers={"Authorization": f"Bearer {WORKER_KEY}"},
    json={"proof_hash": "sha256_of_output"})

# Orchestrator confirms, payment releases (1% fee deducted)
requests.post(f"https://escrow.purpleflea.com/v1/confirm/{escrow['escrow_id']}",
    headers={"Authorization": f"Bearer {ORCHESTRATOR_KEY}"},
    json={"proof_hash": "sha256_of_output"})
Enter fullscreen mode Exit fullscreen mode

Quick Reference

Framework Method Setup time
Claude MCP JSON config 2 min
LangChain MCP adapters or @tool 5 min
CrewAI @tool decorator 5 min
AutoGen function_map 10 min
LlamaIndex FunctionTool 5 min

Get started: faucet.purpleflea.com — free credits, no deposit

Full platform: purpleflea.com

MCP on Smithery: faucet | escrow

Research paper on emergent agent economics: doi.org/10.5281/zenodo.18808440

Top comments (0)