Only scenarios set to On-Demand appear as tools, which makes the Integrator UI the place where you decide what an assistant may run — no allowlist to maintain here. The API key scopes are read and run only, so the assistant can trigger workflows but not rewrite them. This is a fork of Make's original server, maintained by Boost.space and marked legacy.
A bridge between an AI assistant and Boost.space Integrator. It connects to your account, finds the scenarios set to On-Demand scheduling, and offers each one as a callable tool with its inputs described.
- Every On-Demand scenario in your team exposed as a tool, discovered automatically rather than configured by hand
- Input parameters parsed and resolved, so the assistant sees meaningful parameter descriptions instead of raw field names
- Scenario output returned as structured JSON, which the assistant can read and act on
- Automation logic that stays in Integrator — you expose it to the model without rebuilding it
The npm package @boostspace/integrator-mcp-server, version 0.1.4, run as `npx -y @boostspace/integrator-mcp-server`. Two environment variables: `INTEGRATOR_API_KEY`, generated in your Integrator profile with the `scenarios:read` and `scenarios:run` scopes, and `INTEGRATOR_TEAM`, the team id from the URL of your Team page. Node.js is the only local dependency.
One command plus a key — npx -y @boostspace/integrator-mcp-server, then supply credentials