Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to help you build AI agents that turn intent into action.
| 1 | --- |
| 2 | title: Build a Python research agent with MCP |
| 3 | description: Read a public Hacker News profile with CrewAI or LangChain through a Composio session's MCP endpoint. |
| 4 | keywords: [python, mcp, crewai, langchain, hackernews, research agent] |
| 5 | gallery: |
| 6 | categories: [General agents] |
| 7 | logos: [hackernews] |
| 8 | featured: true |
| 9 | order: 7 |
| 10 | --- |
| 11 | |
| 12 | {/* code-example: single-language-page reason="Python-only example project using CrewAI and LangChain Python MCP clients" */} |
| 13 | |
| 14 | Build an agent that reads a Hacker News profile and reports its current karma and bio. Choose CrewAI or LangChain below. Both agents use a Composio session that exposes one public, read-only tool over MCP, so you don't need to connect a Hacker News account. |
| 15 | |
| 16 | The session supplies the MCP URL and authentication headers. Your framework handles tool discovery and execution through that endpoint. |
| 17 | |
| 18 | ## Set up your environment |
| 19 | |
| 20 | Use Python 3.12 and create a virtual environment: |
| 21 | |
| 22 | ```bash |
| 23 | mkdir python-mcp-agent |
| 24 | cd python-mcp-agent |
| 25 | python3.12 -m venv .venv |
| 26 | source .venv/bin/activate |
| 27 | ``` |
| 28 | |
| 29 | Set your [Composio project API key](https://dashboard.composio.dev/~/project/settings/api-keys?utm_source=docs&utm_medium=content&utm_campaign=examples-python-mcp-agents) and [OpenAI API key](https://platform.openai.com/api-keys) in this shell: |
| 30 | |
| 31 | ```bash |
| 32 | export COMPOSIO_API_KEY="your-composio-api-key" |
| 33 | export OPENAI_API_KEY="your-openai-api-key" |
| 34 | export COMPOSIO_EXAMPLES_USER_ID="mcp-research-demo" |
| 35 | ``` |
| 36 | |
| 37 | `COMPOSIO_EXAMPLES_USER_ID` identifies this demo's Composio user. In your application, use your signed-in user's ID. |
| 38 | |
| 39 | ## Run the agent |
| 40 | |
| 41 | Choose one framework and install its dependencies. Each script creates the session, runs the lookup, and deletes the session when the run finishes or raises an error. |
| 42 | |
| 43 | {/* code-example: single-language reason="CrewAI vs LangChain framework picker; both are Python-only MCP agent scripts" */} |
| 44 | <Tabs items={['CrewAI', 'LangChain']}> |
| 45 | <Tab value="CrewAI"> |
| 46 | |
| 47 | Install Composio, CrewAI, and its MCP dependency: |
| 48 | |
| 49 | ```bash |
| 50 | python -m pip install "composio>=0.17.1" crewai mcp |
| 51 | ``` |
| 52 | |
| 53 | Save this as `crewai_agent.py`. CrewAI's [`MCPServerHTTP`](https://docs.crewai.com/en/mcp/overview) accepts the session's URL and headers through the agent's `mcps` list. |
| 54 | |
| 55 | ```python title="crewai_agent.py" |
| 56 | import os |
| 57 | |
| 58 | from composio import Composio, SESSION_PRESET_DIRECT_TOOLS |
| 59 | from crewai import Agent, Crew, Task |
| 60 | from crewai.mcp import MCPServerHTTP |
| 61 | |
| 62 | composio = Composio(api_key=os.environ["COMPOSIO_API_KEY"]) |
| 63 | session = composio.create( |
| 64 | user_id=os.environ["COMPOSIO_EXAMPLES_USER_ID"], |
| 65 | toolkits=["hackernews"], |
| 66 | tools={"hackernews": {"enable": ["HACKERNEWS_GET_USER"]}}, |
| 67 | session_preset=SESSION_PRESET_DIRECT_TOOLS, |
| 68 | mcp=True, |
| 69 | ) |
| 70 | |
| 71 | try: |
| 72 | agent = Agent( |
| 73 | role="Hacker News researcher", |
| 74 | goal="Report public profile information from live tool results.", |
| 75 | backstory="You check the source before writing a short profile.", |
| 76 | llm="gpt-5.2", |
| 77 | mcps=[ |
| 78 | MCPServerHTTP( |
| 79 | url=session.mcp.url, |
| 80 | headers=session.mcp.headers, |
| 81 | streamable=True, |
| 82 | ) |
| 83 | ], |
| 84 | ) |
| 85 | task = Task( |
| 86 | description=( |
| 87 | "Look up the Hacker News user pg with the available tool. " |
| 88 | "Report their current karma and summarize their bio if present. " |
| 89 | "Include https://news.ycombinator.com/user?id=pg as the source. " |
| 90 | "If the lookup fails, report the error instead of guessing." |
| 91 | ), |
| 92 | expected_output="A short profile with username, karma, bio, and source URL.", |
| 93 | agent=agent, |
| 94 | ) |
| 95 | result = Crew(agents=[agent], tasks=[task]).kickoff() |
| 96 | print(result.raw) |
| 97 | finally: |
| 98 | session.delete() |
| 99 | ``` |
| 100 | |
| 101 | Run it: |
| 102 | |
| 103 | ```bash |
| 104 | python crewai_agent.py |
| 105 | ``` |
| 106 | |
| 107 | The repository's [CrewAI MCP example](https://github.com/ComposioHQ/composio/blob/next/python/examples/tool_router/crewai_agent.py) uses the same connection pattern to summarize Gmail. That variant needs a connected Gmail account. |
| 108 | |
| 109 | </Tab> |
| 110 | <Tab value="LangChain"> |
| 111 | |
| 112 | Install Composio, LangChain's MCP extra, and the OpenAI integration: |
| 113 | |
| 114 | ```bash |
| 115 | python -m pip install "composio>=0.17.1" "langchain[mcp]>=1.4,<2" langchain-openai |
| 116 | ``` |
| 117 | |
| 118 | <Callout type="info"> |
| 119 | LangChain's [`MCPAdapter`](https://docs.langchain.com/oss/python/langchain/mcp) requires `langchain[mcp]>=1.4.0` and is in beta. It replaces the archived `langchain-mcp-adapters` package. For an existing `MultiServerMCPClient` integration, follow LangChain's [migration guide](https://docs.langchain.com/oss/python/migrate/langchain-mcp-adapters). |
| 120 | </Callout> |
| 121 | |
| 122 | Save this as `langchain_agent.py`. [`StreamableHttpTransport`](https://gofastmcp.com/clients/transports#http-transport) passes the session's authentication headers to the MCP endpoint. |
| 123 | |
| 124 | ```python title="langchain_agent.py" |
| 125 | import asyncio |
| 126 | import os |
| 127 | |
| 128 | from composio import Composio, SESSION_PRESET_DIRECT_TOOLS |
| 129 | from fastmcp.client.transports import StreamableHttpTransport |
| 130 | from langchain.agents import create_agent |
| 131 | from langchain.mcp import MCPAdapter |
| 132 | from langchain_openai import ChatOpenAI |
| 133 | |
| 134 | |
| 135 | async def main(): |
| 136 | composio = Composio(api_key=os.environ["COMPOSIO_API_KEY"]) |
| 137 | session = composio.create( |
| 138 | user_id=os.environ["COMPOSIO_EXAMPLES_USER_ID"], |
| 139 | toolkits=["hackernews"], |
| 140 | tools={"hackernews": {"enable": ["HACKERNEWS_GET_USER"]}}, |
| 141 | session_preset=SESSION_PRESET_DIRECT_TOOLS, |
| 142 | mcp=True, |
| 143 | ) |
| 144 | try: |
| 145 | transport = StreamableHttpTransport( |
| 146 | url=session.mcp.url, |
| 147 | headers=session.mcp.headers, |
| 148 | ) |
| 149 | async with MCPAdapter(transport) as adapter: |
| 150 | tools = await adapter.list_tools() |
| 151 | agent = create_agent( |
| 152 | model=ChatOpenAI(model="gpt-5.2"), |
| 153 | tools=tools, |
| 154 | ) |
| 155 | result = await agent.ainvoke( |
| 156 | { |
| 157 | "messages": [ |
| 158 | { |
| 159 | "role": "user", |
| 160 | "content": ( |
| 161 | "Look up the Hacker News user pg with the available tool. " |
| 162 | "Report their current karma and summarize their bio if present. " |
| 163 | "Include https://news.ycombinator.com/user?id=pg as the source. " |
| 164 | "If the lookup fails, report the error instead of guessing." |
| 165 | ), |
| 166 | } |
| 167 | ] |
| 168 | } |
| 169 | ) |
| 170 | print(result["messages"][-1].content) |
| 171 | finally: |
| 172 | session.delete() |
| 173 | |
| 174 | |
| 175 | if __name__ == "__main__": |
| 176 | asyncio.run(main()) |
| 177 | ``` |
| 178 | |
| 179 | Run it: |
| 180 | |
| 181 | ```bash |
| 182 | python langchain_agent.py |
| 183 | ``` |
| 184 | |
| 185 | The `async with` block closes the MCP connection before `finally` deletes the Composio session. |
| 186 | |
| 187 | The same script is in [`python/examples/tool_router/langchain_agent.py`](https://github.com/ComposioHQ/composio/blob/next/python/examples/tool_router/langchain_agent.py). From a repository checkout, run it with [uv](https://docs.astral.sh/uv/guides/scripts/) to install its declared dependencies in an isolated environment: |
| 188 | |
| 189 | ```bash |
| 190 | uv run --script python/examples/tool_router/langchain_agent.py |
| 191 | ``` |
| 192 | |
| 193 | </Tab> |
| 194 | </Tabs> |
| 195 | |
| 196 | ## Check the result |
| 197 | |
| 198 | The agent prints `pg`'s current karma, a short bio when the profile contains one, and the Hacker News profile URL. Values and wording vary between runs. Check that the response cites `https://news.ycombinator.com/user?id=pg` and uses the tool's result. |
| 199 | |
| 200 | The [direct tools preset](/docs/configuring-sessions#direct-tools-preset) exposes `HACKERNEWS_GET_USER` directly. The agent doesn't need to search for tools or start an authentication flow. To research another Hacker News user, change `pg` in the prompt and source URL. |
| 201 | |
| 202 | If the MCP connection returns an authentication error, check `COMPOSIO_API_KEY` and pass both `session.mcp.url` and `session.mcp.headers`. Re-run the script to create a fresh session; the previous run's session was deleted. |
| 203 | |
| 204 | ## Use the same pattern in your application |
| 205 | |
| 206 | Keep the [session](/docs/how-composio-works) for as long as the conversation needs it, then delete it. To use a toolkit that needs an account, [authorize that toolkit](/docs/authentication/manually-authenticating) before running the agent. |
| 207 | |
| 208 | For SDK modifiers or custom tools that run in your process, use the [CrewAI provider](/docs/providers/crewai) or [LangChain provider](/docs/providers/langchain). Those features don't run through a hosted [MCP session](/docs/sessions-via-mcp#trade-offs). |