Hey Makers
We’re excited to introduce MCP Toolboxes: a new way to safely power AI agents and AI apps with purpose-built Make tools.
If you’re empowering AI clients like Claude, Cursor, ChatGPT, or custom AI apps, MCP Toolboxes give you complete managed control so that they reliably execute exactly the Make scenarios you choose.
What can you do with MCP Toolboxes?
Create dedicated MCP servers with a subset of scenarios
Give agents only the tools they need (better security + reliability)
Power different AI agents or apps with different toolsets
Manage access using dedicated tokens/keys
Get basic visibility into tool usage per toolbox
Why it matters
Giving agents access to everything can make them:
- Less reliable
- More expensive to run
- Risky in multi-customer or multi-team setups
MCP Toolboxes help you build safer, more predictable, and more scalable agentic systems — with Make as the orchestration layer.
How it works
Create a toolbox at the team level, assign the scenarios you want to expose, and connect your AI client using the toolbox’s MCP URL and token. That’s it.
Who is this for?
Advanced automation builders
Teams building AI agents outside of Make
Platform builders embedding agents for their own customers
Learn more
Check out the docs and demo to see MCP Toolboxes in action.
Already building agents with Make? Tell us how you’re planning to use toolboxes — we’d love to learn from your setup!