MCP 服务器

Superglue MCP

A service that enables agents to build deterministic workflows across apps, databases and APIs using natural language, handling schema mapping, drift detection, and execution of integration pipelines automatically.

README

<h2 align="center">Integrate & Orchestrate APIs with natural language.</h2> <div align="center">

</div> <h3 align="center"> Now live: let agents build deterministic workflows across apps, databases and APIs using the superglue MCP<br> Let's glue.<br>

Read the docs 🍯🍯🍯</h3>

what is superglue?

superglue orchestrates APIs from natural language. Tell it what you want to do in your CRM, ERP and co. and superglue builds, runs and executes the integration pipelines automatically. Comes with automated schema mapping, drift detection, retries and remappings so your API workflows keep running no matter what. superglue makes agents reliable in prod by letting them build deterministic workflows across any SaaS app, API and data source. Use the superglue MCP instead of hard-coding tools and let your agent use APIs the way they want to, not the way they were written.

Lightweight proxy: point it at any REST / GraphQL / SQL / postgres / file endpoint.
LLM‑assisted mapping during config; cached Javascript transforms at runtime (no LLM latency).
Self‑heals schema drift: when the upstream API or schema changes, superglue regenerates the transform automatically, and keeps the pipeline running.
Security‑first: zero data stored; run fully on‑prem or use our hosted version.

quick start

hosted version

Run on our cloud-hosted version
Install the superglue js/ts client:

npm install @superglue/client

Configure your first api call:

import { SuperglueClient } from "@superglue/client";

const superglue = new SuperglueClient({
  apiKey: "************"
});

const workflowResult = await superglue.executeWorkflow({
  // input can be an ID of a pre-saved workflow or a WorkflowInput object
    workflow: {
      id: "myTodoUserWorkflow",
      steps: [
        {
          id: "fetchTodos", // Unique ID for this step
          apiConfig: {
            id: "jsonplaceholderTodos",
            urlHost: "https://jsonplaceholder.typicode.com",
            urlPath: "/todos",
            method: HttpMethod.GET,
            instruction: "Fetch a list of todos. We only need the first one for this example.",
          },
        },
        {
          id: "fetchUser",
          apiConfig: {
            id: "jsonplaceholderUsers",
            urlHost: "https://jsonplaceholder.typicode.com",
            urlPath: "/users/<<$.fetchTodos[0].userId>>", // JSONata path parameter for first userId
            method: HttpMethod.GET,
            instruction: "Fetch user details by user ID for the first todo."
          },
        },
      ],
      // Transform the results of the steps into the final desired output. If not given, this will be generated from the reponse schema
      finalTransform: "$",
      responseSchema: { // define the expected final output structure
        type: "object",
        description: "first todo",
        properties: {
            todoTitle: { type: "string" },
            userName: { type: "string" }
        }
      }
  },
  // `payload` could be used to pass initial data to the first step if needed. E.g. IDs to fetch, filters, etc. In short, things that can change across calls.
  // payload: { userId: 1 },
  // `credentials` can be used to authenticate requests. They need to be referenced in the api config (e.g. "headers": {"Authorization": "Bearer <<hubspot_api_key>>"})
  // credentials: { hubspot_api_key: "pa_xxx" },      
});
console.log(JSON.stringify(workflowResult, null, 2));

what people build with superglue

Voice assistants: reliably map intent to tool usage
Extended GPT: offer more data sources and a whitelabel agent builder inside your internal GPT
Extend AI assistant/co-pilot: offer more actions than search
Ship connectors 10x faster, without the maintenance overhead
Simple interface for legacy API pipelines
CMS or cloud migration
Transforming SQL queries into Rest API calls
And many more...

key features

API Proxy: Configure APIs and intercept responses in real-time with minimal added latency
LLM-Powered Data Mapping: Automatically generate data transformations using large language models
Schema Validation: Ensure data compliance with your specified schemas
File Processing: Handle various file formats (CSV, JSON, XML) with automatic decompression
Flexible Authentication: Support for various auth methods including header auth, api keys, oauth, and more
Smart Pagination: Handle different pagination styles automatically
Caching & Retry Logic: Built-in caching and configurable retry strategies