---
id: arcade-mcp
version: "1.15.2"
license: MIT License Copyright (c) 2025, Arcade AI Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in… (full text in the JSON record)
license_treatment: permissive
maintenance: active
---

# arcade-mcp — Arcade.dev - Tool Calling platform for Agents

License: permissive · Maintenance: active · Downloads: 96.4K/mo

## What it is and what it does

Arcade-mcp is a Python framework for building MCP (Model Context Protocol) servers—the standardized interface that lets LLM agents and clients call tools. It abstracts away the protocol complexity behind a decorator API, so you define tools as simple Python functions and the framework handles MCP serialization, transport (stdio or HTTP), and client negotiation.

The framework's main value is authorized tool calling: you declare which OAuth scopes or API keys a tool needs (e.g., `requires_auth=GitHub(scopes=["repo"])`), and Arcade handles the OAuth flow, token storage, refresh, and per-call injection—all without exposing secrets to the LLM or client. It ships with 22 pre-built OAuth helpers for common providers and a generic OAuth2 class for custom APIs. Beyond tools, it supports MCP resources, prompts, sampling, progress reporting, and logging. You can run servers standalone over stdio or HTTP, or deploy to Arcade Cloud for managed hosting and observability.

Use it for:

- Build an internal API tool that your LLM agent can call with OAuth-protected access to your company's systems.
- Scaffold a GitHub integration tool that lists or modifies repos on behalf of an authenticated user, with token lifecycle handled automatically.
- Create a multi-tool MCP server combining custom business logic with pre-built OAuth helpers (Slack, Notion, etc.) for a unified agent interface.
- Deploy a tool server to Arcade Cloud and expose it to multiple LLM clients (Claude, Cursor, VS Code) without managing infrastructure.
- Test tool-call accuracy against real LLMs using the `arcade evals` command before production deployment.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Framework for building Model Context Protocol (MCP) servers and tools in Python, with built-in support for OAuth-authorized tool calling and integration with LLM agents.

Yes. Active, production-stable framework with low install friction, permissive MIT license, and no known vulnerabilities. Install if you need to build MCP servers with OAuth-authorized tool calling for LLM agents. Skip if you only consume pre-built tools from Arcade.dev's catalog or don't need agent tool integration.

## Install

pip install arcade-mcp
uv add arcade-mcp
poetry add arcade-mcp

## Installing arcade-mcp

Before you install: Low friction install via wheel distribution. Active maintenance with release 4 days old. Supports Python 3.10–3.13 and marked Production/Stable.

License in practice: MIT License permits commercial and private use, modification, and distribution with minimal restrictions—only requiring license and copyright notice retention.

Quickstart:

# Install
pip install arcade-mcp

# Minimal server
from arcade_mcp_server import MCPApp
from typing import Annotated

app = MCPApp(name="my_server", version="1.0.0")

@app.tool
def greet(name: Annotated[str, "Name to greet"]) -> str:
    """Greet a person by name."""
    return f"Hello, {name}!"

if __name__ == "__main__":
    app.run(transport="stdio")

Requires Python 3.10 or later.

Verify before relying:

- Whether the 22 pre-built OAuth helper classes (GitHub, Slack, etc.) cover the specific providers your integration needs.
- Performance characteristics and rate-limiting behavior when hosting multiple tools on Arcade Cloud.
- Whether Arcade Cloud deployment incurs hosting costs or usage fees beyond the open-source framework itself.

## Package facts

- License: MIT License Copyright (c) 2025, Arcade AI Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in… (full text in the JSON record) (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 96.4K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

MCP server framework Python, model context protocol tools, LLM tool calling OAuth, agent framework integration, authorized API access for agents, MCP decorator API, tool server scaffolding, llm-agents, oauth-integration, mcp-protocol

[View on SkillFed](https://skillfed.io/packages/arcade-mcp) · [View on PyPI](https://pypi.org/project/arcade-mcp/)