AI Agent Access (MCP)
Connect Claude, Cursor, ChatGPT and Gemini to your maintenance data Prevate exposes 152 industrial tools via the Model Context Protocol (MCP) so your team can query work orders, spares, PM schedules and procurement in plain English — with the same org RBAC and row-level security as the web app.
What is MCP?
The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data. Prevate runs an MCP server so Claude, Cursor, ChatGPT and Gemini can query and act on your maintenance data securely.
Which AI clients are supported?
Claude (claude.ai web via OAuth connector, plus Claude Desktop), Cursor, ChatGPT and Google Gemini. Each connects to the Prevate MCP server at minesite-mcp.fly.dev.
How do I connect?
Sign in to Prevate, open Account Settings → MCP Connection, and generate a token or follow the setup kit for Claude Desktop and Cursor. Claude web uses OAuth — add a custom connector in Claude settings and sign in when prompted.
Is my org data isolated?
Yes. Every MCP call uses your Supabase user JWT. Org isolation uses the same RLS and mine-site filters as the web app. The MCP server never stores a service role for tenant data.
How is this different from the AI Voice Companion?
Voice is for hands-free field use on the iPhone. MCP is for desktop and agent clients — power users, planners and automation workflows that live in Claude, Cursor, ChatGPT or Gemini.
Does Prevate also offer a REST API?
Yes. For programmatic integrations and webhooks, see the REST API & Webhooks integration page. MCP is the best path when you want AI agents to work with maintenance data in natural language.
Learn more at https://prevate.ai. Pricing: https://prevate.ai/pricing. Sitemap: https://prevate.ai/sitemap.xml. LLM summary: https://prevate.ai/llms.txt.