{"categories":[{"label":"Application Frameworks","url":"https://skillfed.io/packages/category/software-development-libraries-application-frameworks/5"}],"enrichment":{"capability":"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.","skillfed_tags":["llm-agents","oauth-integration","mcp-protocol"],"use_cases":["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."],"what_it_does":"Arcade-mcp is a Python framework for building MCP (Model Context Protocol) servers\u2014the 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.\n\nThe 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\u2014all 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.","worth_installing":"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."},"id":"arcade-mcp","links":{"html":"https://skillfed.io/packages/arcade-mcp","md":"https://skillfed.io/packages/arcade-mcp.md","pypi":"https://pypi.org/project/arcade-mcp/"},"maintenance":{"status":"active"},"meta":{"latest_release":"2026-08-10","license_spdx":null,"license_treatment":"permissive","name":"arcade-mcp","python_support":"supports_current","summary":"Arcade.dev - Tool Calling platform for Agents"},"popularity":{"monthly_downloads":96376,"position":13216,"tier":"top_15000"},"security":{"n_vulnerabilities":0},"version":"1.15.2"}