Announcing MCPHero Python Package - map MCP tools to OpenAI tools

Source: Dev.to

mcphero: Use MCP servers with native OpenAI clients

I released a small Python package that maps MCP servers to native OpenAI clients.

• PyPI:
• GitHub:

The problem

• Native openai / gemini clients don’t support MCP.
• Because of that, many projects don’t use MCP at all.
• Or projects have to write a lot of custom mapping code.

What mcphero does

• Converts MCP tools into OpenAI‑compatible tools/functions.
• Sends LLM tool‑call results back to the MCP server for execution.
• Returns updated message history.
• Provides two‑way mapping in ~2 lines of code.

Minimal example

adapter = MCPToolAdapterOpenAI(
    "https://api.mcphero.app/mcp/your-server-id/mcp"
)
client = OpenAI()

tools = await adapter.get_tool_definitions()

response = client.chat.completions.create(
    model="gpt-5",
    messages=[{"role": "user", "content": "What's the weather in London?"}],
    tools=tools,
)

tool_calls = response.choices[0].message
result = await adapter.process_tool_calls(tool_calls)

• Without get_tool_definitions(), MCP tools can’t be wired into the OpenAI tools argument.
• Without process_tool_calls(), OpenAI tool_calls can’t be wired into the MCP server.

Feedback is welcome.

If you are looking for a managed solution for MCP servers, try mcphero.app – AI‑generated MCP servers built for developers.