Build your own server (Python)

Build a small Python MCP server that wraps the axiom.ai REST API. The result is a tool any MCP-aware client (Claude Desktop, Claude Code, Cursor) can call to trigger automations and check their status from inside a conversation.

Before you begin

You need:

• Python 3.10 or later.
• uv (recommended) or pip for dependency management.
• An axiom.ai API key. See Authentication.

Step 1: install dependencies

uv pip install "mcp[cli]" httpx

Or with pip:

pip install "mcp[cli]" httpx

Step 2: write the server

Create a file called axiom_mcp.py:

import os
import httpx
from mcp.server.fastmcp import FastMCP

mcp = FastMCP("axiom")
AXIOM_BASE = "https://lar.axiom.ai/api/v3"
KEY = os.environ["AXIOM_API_KEY"]

@mcp.tool()
async def trigger_automation(name: str, data: list | None = None) -> dict:
    """Trigger an axiom.ai cloud automation by exact name.

    Optional `data` is a 3-D array of [tables][rows][cells] passed as
    the automation's input.
    """
    body = {"key": KEY, "name": name}
    if data is not None:
        body["data"] = data
    async with httpx.AsyncClient(timeout=30) as c:
        r = await c.post(f"{AXIOM_BASE}/trigger", json=body)
        return r.json()

@mcp.tool()
async def run_status(run_id: str) -> dict:
    """Check the status of a triggered axiom.ai automation run."""
    async with httpx.AsyncClient(timeout=30) as c:
        r = await c.post(
            f"{AXIOM_BASE}/run-data",
            json={"key": KEY, "run_id": run_id},
        )
        return r.json()

if __name__ == "__main__":
    mcp.run()  # stdio transport by default

That's the whole server. Tool descriptions in the docstrings are what the LLM sees when deciding whether to call a tool, write them clearly.

Step 3: register it with a client

See Register with Claude clients for the full configuration snippets. The shortest version, for Claude Desktop:

{
  "mcpServers": {
    "axiom": {
      "command": "uv",
      "args": ["run", "/absolute/path/to/axiom_mcp.py"],
      "env": { "AXIOM_API_KEY": "your-api-key-here" }
    }
  }
}

Restart the client. Your trigger_automation and run_status tools appear and the LLM can call them in conversation.

Step 4: extend it

The minimal server above is intentionally tiny. Real custom servers usually add:

• stop_run to cancel misbehaving runs.
• remaining_runtime so the LLM can check quota before triggering expensive runs.
• Domain-specific wrappers, for example scrape_amazon_asin(asin: str) that internally calls trigger_automation("Amazon ASIN template", data=[[[asin]]]) and waits for completion. This is where the real value is, the LLM gets a clean, typed tool instead of having to learn each automation's input shape.
• Polling and completion handling, so the tool returns a finished result rather than a run_id the LLM has to poll itself.
• Step-function tools, mirroring the Step API if you want the LLM to drive a session step by step instead of just triggering pre-built automations.

A typical extended server is 100 to 200 lines and covers the most common workflows for the team using it.

Notes

• Treat your AXIOM_API_KEY like any other secret. Don't commit the env value into the config you check into version control, use a secrets manager or a .env file the client loads.
• The server runs as a child process of the MCP client. If it crashes, the client restarts it on the next tool call, so transient errors are usually fine.
• For production, consider switching from stdio to streamable HTTP transport so the server can run remotely. See the Anthropic build-a-server guide.