arcade-serve

What it is and what it does

Arcade Serve is a hosting framework built on FastAPI and the Model Context Protocol that lets you run Arcade tools as networked services. It handles the plumbing: worker lifecycle, toolkit registration, authentication, and observability through OpenTelemetry. The package exposes tools via HTTP (FastAPI) or stdio (MCP), complete with request tracing and monitoring hooks.

The package is designed for developers building tool-serving infrastructure—either standalone FastAPI applications or MCP servers that integrate with language models. It abstracts away the common patterns of authentication, toolkit loading, and runtime management, letting you focus on which tools to expose and how to configure them.

Use it for:

• Host Arcade tools as a FastAPI microservice with secret-based authentication.
• Expose tools via Model Context Protocol for integration with language model clients.
• Build a multi-worker deployment where different workers host different tool subsets.
• Add observability to tool execution by leveraging built-in OpenTelemetry instrumentation.
• Manage tool lifecycle and registration without writing boilerplate authentication or routing code.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Arcade Serve provides FastAPI and Model Context Protocol server infrastructure for hosting Arcade tools and workers with built-in authentication, observability, and toolkit registration.

Yes, if you are building or deploying Arcade tools as services. The package is actively maintained, has no known vulnerabilities, and eliminates boilerplate for FastAPI and MCP hosting. Install it when you need to expose toolkits over HTTP or MCP; skip it if you are only using tools locally or in a single-process context.

Install

arcade-serve on PyPI

pip install arcade-serve

uv add arcade-serve

poetry add arcade-serve

Installing arcade-serve

Before you install

Low install friction with a pure-Python wheel. Marked active with a release 1 day old. Depends on FastAPI, uvicorn, aiohttp, and OpenTelemetry instrumentation packages—all stable, widely-used libraries.

License in practice

MIT license (permissive). You can use, modify, and distribute this package freely in commercial or private projects with minimal restrictions.

Quickstart

pip install arcade-serve

from fastapi import FastAPI
from arcade_serve.fastapi import FastAPIWorker

app = FastAPI()
worker = FastAPIWorker(app, secret=os.environ.get("ARCADE_WORKER_SECRET"))
worker.register_toolkit(...)

Requires Python 3.10 or later. Expects ARCADE_WORKER_SECRET environment variable for worker authentication.

Verify before relying

• Whether OpenTelemetry instrumentation is automatically active or requires explicit configuration.
• Performance characteristics and throughput limits for concurrent tool execution.
• Compatibility with custom authentication schemes beyond environment-variable-based secrets.
• How toolkit registration works and what toolkit objects the register_toolkit method accepts.

Package facts

Evidence: arcade_serve-3.4.2-py3-none-any.whl

Tags

More Application Frameworks packages