MCP Servers · ArcadeAI

arcade-mcp

arcade-mcp is an open-source Python framework for building Model Context Protocol (MCP) servers that extend AI agents with custom tools and integrations. It provides decorator-based APIs for tool development, built-in OAuth/secret management, and deployment options via Arcade Cloud or standalone hosting.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/ArcadeAI/arcade-mcp

952

GitHub stars

101

Forks

Python

Primary language

MIT

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	ArcadeAI/arcade-mcp
Owner	ArcadeAI
Primary language	Python
License	MIT — OSI-approved
Stars	952
Forks	101
Open issues	20
Latest release	Unknown
Last updated	2026-07-06
Source	https://github.com/ArcadeAI/arcade-mcp

What arcade-mcp is

Python framework implementing the Model Context Protocol specification with support for tools, resources, prompts, sampling, and authorized tool calling. Features decorator-driven development, OAuth provider helpers for 22+ services, MCP transport options (stdio, HTTP+SSE), and optional cloud deployment with Arcade.

Quickstart

Get the arcade-mcp source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/ArcadeAI/arcade-mcp.git2cd arcade-mcp3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Custom internal API integrations for agents

Expose internal APIs, microservices, or domain-specific systems as MCP tools without modifying core agent infrastructure. Handle OAuth and secrets centrally while agents remain vendor-neutral.

Rapid prototyping of agent capabilities

Use decorator-based tool definitions and built-in OAuth helpers to quickly scaffold and test custom tool functionality against real LLMs via arcade evals before production deployment.

Multi-LLM and multi-client tool distribution

Build once, run anywhere: one MCP server works with Claude, Cursor, VS Code, and any MCP-compatible client or LLM without rewriting integrations.

Implementation considerations

Requires Python 3.10+ and uv package manager; evaluate existing team tooling and CI/CD compatibility.
OAuth-scoped tool calling works best with Arcade Cloud; standalone deployment requires manual token management and Resource Server Auth implementation.
No latest release tag in repo; pin to specific commits or PyPI versions and monitor for breaking changes during development.
Tool-call accuracy testing via arcade evals depends on LLM availability and cost; budget for eval infrastructure.
Decorator-based design is intuitive but couples tool logic to MCP framework; refactoring to standalone logic later may require rework.

When to avoid it — and what to weigh

You need non-Python environments — arcade-mcp is Python-only. Non-Python teams or polyglot deployments will require language-specific MCP implementations or wrapper layers.
You require guaranteed long-term backwards compatibility — Project created April 2024 with active development but no versioned releases tagged. API stability and breaking change frequency are unknown; production use without version pinning carries risk.
You need advanced security certifications (SOC2, FedRAMP, HIPAA) — No evidence of formal security certifications or audit history. Authorized tool calling delegates token/secret management to Arcade Cloud, requiring trust in their infrastructure.
You want fully offline, air-gapped deployments — Standalone mode is supported but OAuth flows and optional Arcade Cloud integration assume internet connectivity and external service dependencies.

License & commercial use

MIT License (OSI-approved, permissive). Allows commercial use, modification, and distribution without attribution requirement, subject only to license and copyright notice inclusion.

MIT is a permissive OSI license. Commercial use is permitted. However, no explicit commercial support SLA, warranty disclaimer, or indemnification is evident from the project. Any production use should include independent security review, SLA negotiation with Arcade (if using Cloud), and vendor risk assessment.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Adequate
License clarity	Clear
Deployment complexity	Low
DEV.co fit	Strong
Assessment confidence	High

Security considerations

OAuth token and API key injection via Arcade Cloud encrypts secrets at rest but centralizes credential management; requires trust in Arcade's infrastructure. Standalone deployments must implement token protection independently. No public vulnerability history, security audit, or formal threat model documented. SECURITY.md file exists but content not provided. Recommend security review before handling sensitive credentials.

Alternatives to consider

Anthropic MCP SDK (TypeScript/Python reference implementation)

Official MCP specification reference; lower-level control but less out-of-box tooling and no built-in OAuth. Suitable for teams seeking minimal abstraction or non-Python environments.

LangChain Tool/Agent framework

Mature ecosystem with broader agent abstractions and multi-LLM support. MCP integration is newer; better if you're already in LangChain ecosystem but less MCP-native.

Custom MCP server (hand-rolled)

Maximum control and minimal dependencies. Suitable for simple, single-purpose tools or organizations avoiding framework lock-in but requires more boilerplate.

Software development agency

Build on arcade-mcp with DEV.co software developers

Start building custom tools with arcade-mcp. Use the CLI to scaffold a new server, integrate internal APIs, and deploy to Arcade Cloud or run standalone.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

mcp-agent mcp mcp-use

Related on DEV.co

Explore the category and the services that help you build with it.

MCP Servers AI Application Development API Development Cloud Deployment Talk to DEV.co

arcade-mcp FAQ

Does arcade-mcp work with non-Arcade Cloud deployments?

Yes. Standalone mode runs on any machine over stdio or HTTP. You supply your own token management, secret injection, and (for HTTP) endpoint authentication via Resource Server Auth.

Can I use arcade-mcp tools with GPT-4, Claude, or other LLMs?

Yes. MCP is vendor-neutral. Tools work with any MCP-compatible client and LLM. Authorized OAuth flows are easiest via Arcade Cloud; standalone requires manual token handling.

What happens if Arcade Cloud goes down?

Cloud-hosted tools and OAuth flows become unavailable. Standalone servers continue operating independently. Hybrid strategies (some tools standalone, some Cloud-hosted) are possible.

Is there a production SLA or support contract?

Not stated in the data. arcade-mcp is open-source MIT; production support and SLAs depend on Arcade company offerings, not the framework. Review Arcade's commercial terms separately.

Software developers & web developers for hire

Need help beyond evaluating arcade-mcp? DEV.co is a software development agency offering software development services and web development for teams of every size. Our software developers and web developers build custom software, web applications, APIs, and mcp servers integrations — and maintain them long-term.

Ready to extend your AI agents?

Start building custom tools with arcade-mcp. Use the CLI to scaffold a new server, integrate internal APIs, and deploy to Arcade Cloud or run standalone.

Talk to DEV.co Schedule a Consultation