Arcade

Arcade is an MCP-native tool execution, hosting, and governance platform built by Arcade AI, Inc that enables AI agents to connect to and take actions in external services through pre-built, authenticated integrations and Arcade-hosted MCP servers. The platform is designed around the Model Context Protocol from the ground up, providing agents with standardised tool interfaces for services including GitHub, Google Workspace, Salesforce, Slack, and others without requiring developers to manage OAuth flows or credential storage themselves. Every pricing tier, including the free Hobby plan, includes MCP compatibility and Arcade-hosted MCP server capacity. Pricing is usage-based, verified live on arcade.dev/pricing on May 6, 2026: Hobby (Free) at $0 includes 100 user challenges per month (authentication challenges for new permissions), 1,000 standard tool executions per month, 50 pro tool executions per month for advanced capabilities, 1 Arcade-hosted MCP server, and 24/7 community support via GitHub. Growth at $25 per month includes 600 user challenges (then $0.05 each), 2,000 standard tool executions (then $0.01 each), 100 pro tool executions (then $0.50 each), Arcade-hosted MCP servers at $0.05 per hour, and email support with Service Level Agreement. Enterprise is custom-priced and includes all Growth features, custom pricing across usage metrics, dedicated tenant isolation, audit logs and compliance reporting, role-based access control (RBAC), SSO and SAML authentication, and a dedicated account representative. A startup programme is available for companies under 100 employees, nonprofits, and educational institutions. Setup takes approximately 15 minutes from account creation to first authenticated tool execution via the Arcade SDK. User challenges are the number of times an end-user is challenged for new permissions when connecting a service -- after the initial challenge, subsequent tool calls to the same service execute without re-authentication. Standard tools have lower infrastructure costs; Pro tools are advanced capabilities with higher infrastructure overhead. Autonomous rate is very high: after initial OAuth setup, Arcade executes tool calls on behalf of AI agents without per-action user intervention, including scheduled executions. Security: Arcade AI, Inc is a US corporation. The privacy policy (effective March 2, 2026) confirms AWS infrastructure for data storage, encryption safeguards, access controls, and GDPR rights coverage for EU residents. No SOC-2 certification or ISO 27001 is mentioned in the privacy policy or on any public security documentation page. Enterprise tier includes audit logs and compliance reporting. Limits: free tier''s 1,000 standard tool executions can exhaust quickly for agents running multi-step workflows; Pro tool executions at $0.50 each on overage create significant cost at scale for agents using advanced capabilities; no SOC-2 or ISO 27001 confirmation from public documentation -- Enterprise compliance documentation requires a sales engagement; the platform is developer-facing and requires SDK integration. TCO at Growth for a team at $25/mo is $300/year base plus usage; additional MCP server hours at $0.05/hr equate to $36/year per always-on server.