Use Bazaar MCP Tools with AutoGen

Prerequisites

• Python 3.10+
• A Bazaar API key (get one here)
• Funded Bazaar balance (see billing setup)

1. Install Dependencies

pip install autogen-agentchat autogen-ext[mcp]

AutoGen's autogen-ext[mcp] package includes McpWorkbench, which bridges MCP servers into AutoGen's tool system.

2. Set Environment Variables

export NOUI_API_KEY="bz_your_key_here"
export OPENAI_API_KEY="sk-..."   # or Azure OpenAI credentials

3. Connect AutoGen to Bazaar

Use McpWorkbench to connect to Bazaar's SSE endpoint. All discovered tools are automatically available to your agents:

import os
import asyncio
from autogen_agentchat.agents import AssistantAgent
from autogen_agentchat.ui import Console
from autogen_ext.tools.mcp import McpWorkbench, SseServerParams
from autogen_ext.models.openai import OpenAIChatCompletionClient

API_KEY = os.environ["NOUI_API_KEY"]

# Configure the Bazaar MCP connection
bazaar_params = SseServerParams(
    url="https://noui.bot/api/mcp/sse",
    headers={"Authorization": f"Bearer {API_KEY}"},
)


async def main():
    # Connect to Bazaar and discover tools
    async with McpWorkbench(bazaar_params) as workbench:
        tools = workbench.tools
        print(f"Loaded {len(tools)} Bazaar tools")

        # Create an AutoGen agent with Bazaar tools
        model_client = OpenAIChatCompletionClient(model="gpt-4o")

        agent = AssistantAgent(
            name="bazaar_agent",
            model_client=model_client,
            tools=tools,
            system_message=(
                "You are a helpful agent with access to "
                "Bazaar MCP tools. Use them to answer questions."
            ),
        )

        # Run a task
        result = await Console(
            agent.run_stream(
                task="Check the platform stats on noui.bot"
            )
        )
        print(result)


asyncio.run(main())

4. Multi-Agent Conversations

AutoGen excels at multi-agent setups. Give Bazaar tools to specific agents in a group chat:

from autogen_agentchat.agents import AssistantAgent
from autogen_agentchat.teams import RoundRobinGroupChat
from autogen_agentchat.conditions import TextMentionTermination

async def multi_agent_example():
    async with McpWorkbench(bazaar_params) as workbench:
        model = OpenAIChatCompletionClient(model="gpt-4o")

        # Tool-using agent has Bazaar access
        tool_agent = AssistantAgent(
            name="tool_user",
            model_client=model,
            tools=workbench.tools,
            system_message="You fetch data using Bazaar tools.",
        )

        # Analyst agent has no tools — just reasoning
        analyst = AssistantAgent(
            name="analyst",
            model_client=model,
            system_message=(
                "You analyze data provided by tool_user. "
                "Say TERMINATE when analysis is complete."
            ),
        )

        termination = TextMentionTermination("TERMINATE")
        team = RoundRobinGroupChat(
            [tool_agent, analyst],
            termination_condition=termination,
        )

        result = await Console(
            team.run_stream(
                task="Get noui.bot stats and analyze the trends"
            )
        )
        print(result)

5. How Billing Works

Every tool call through Bazaar is metered transparently:

# Check your balance anytime
curl https://noui.bot/api/v1/bazaar/balance \
  -H "Authorization: Bearer bz_your_key"

# View receipts
curl https://noui.bot/api/v1/bazaar/usage/summary \
  -H "Authorization: Bearer bz_your_key"

6. Production Tips