Arcade is an actions runtime for production AI agents. It helps agents authenticate users, enforce permissions, execute tools, and log actions across business systems.

What is Arcade used for?

Arcade is used when an AI agent needs to take real actions in apps such as email, calendars, CRMs, databases, APIs, or internal tools while keeping authorization and governance under control.

Does Arcade support MCP?

Yes. Arcade positions itself as an MCP runtime and provides agent-optimized tools across MCP servers, along with support for custom tools and existing MCP integrations.

Which agent frameworks work with Arcade?

Arcade documentation mentions integrations with frameworks and clients such as LangChain, CrewAI, OpenAI Agents, Vercel AI, AG2, and MCP clients.

Is Arcade only for enterprises?

No. Arcade can be used by developers building agents, but its strongest fit is production or enterprise-grade agents that need secure authentication, policy enforcement, audit logs, and tool execution.

How much does Arcade cost?

Arcade offers public pricing with a free developer option and paid usage based on platform resources such as auth events and tool calls. Teams should check the official pricing page because usage and bundles can change.

Arcade

MCP runtime for production AI agents with auth, tools, and governance.

Visit Website

What is Arcade?

TL;DR

Arcade is an MCP runtime for production AI agents that handles authentication, authorization, tool execution, governance, and audit logs when agents need to take real actions in external systems and apps.

Connect AI agents to business systems through an MCP runtime
Handle user auth, permissions, policy enforcement, and audit logs
Use agent-optimized tools for reliable external actions

Explore with AI

About

Arcade is an MCP runtime for production AI agents. It sits between agents and the external tools, apps, and business systems they need to access, handling authentication, authorization, tool execution, governance, and audit logs in one runtime layer. The platform is useful for developers and founders building agents that need to take real actions, not just generate text. Agents may need to access Google, Slack, Salesforce, databases, internal tools, or other systems on behalf of real users. Arcade helps ensure those actions respect user permissions, policy rules, and compliance requirements. Arcade's site positions the product around three major agent deployment problems: auth, reliability, and governance. It supports MCP patterns, agent-optimized tools, policy enforcement, auditability, and integration with existing identity providers and frameworks. For FutureStack, Arcade belongs in AI Agents because it solves a hard production problem for agent builders: safely connecting agents to tools and business systems. It is best for technical founders and developers shipping agents that need secure tool access, user-scoped actions, and enterprise-ready governance.

Use Cases

Secure Agent Actions

Connect agents to business systems while enforcing user authentication, permissions, policies, and auditable tool execution.

MCP Tool Infrastructure

Use MCP-compatible tools and runtimes to give agents reliable access to external applications and services.

Enterprise Agent Governance

Apply policy controls and audit trails to agent actions across sensitive business workflows and applications.

Custom Agent Tools

Build and connect custom tools so production agents can safely interact with APIs and internal systems.

Key Features

Connect AI agents to business systems through an MCP runtime
Handle user auth, permissions, policy enforcement, and audit logs
Use agent-optimized tools for reliable external actions
Works with any LLM, framework, identity provider, or MCP client

Featured Tools

Wispr Flow

Voice-First Dictation

We may earn commissions from links to support our work. Learn more.