> ## Documentation Index
> Fetch the complete documentation index at: https://docs.synti.co/llms.txt
> Use this file to discover all available pages before exploring further.

# Connect to Anything

> Reach services that lack a native integration by routing Synti through an API aggregator.

Synti connects to data and tools through three kinds of sources — MCP servers, REST APIs, and local filesystems. But not every service ships its own MCP server. When you need one that doesn't, an **API aggregator** bridges the gap: it exposes hundreds or thousands of apps behind a single MCP endpoint, handling the authentication so you don't have to.

## When you need an aggregator

If the service you want already offers an MCP server or a documented REST API, connect it directly — see [Sources](/sources/overview). Reach for an aggregator when a service has no native integration, or when you'd rather not manage OAuth tokens and refresh flows yourself. The aggregator becomes one more MCP source in your instructions, and its tools appear to the agent like any other.

## Choosing an aggregator

Each platform trades breadth for structure differently. Any of them plugs into Synti the same way.

<CardGroup cols={2}>
  <Card title="Zapier" icon="bolt">
    Thousands of apps and tens of thousands of actions. Broadest coverage for everyday SaaS.
  </Card>

  <Card title="Composio" icon="puzzle-piece">
    Hundreds of tools, plus a universal server that talks to any connected app through a small, fixed tool set.
  </Card>

  <Card title="Make" icon="diagram-project">
    Turns the visual scenarios you build into callable tools with inputs and outputs detected automatically.
  </Card>

  <Card title="Pipedream" icon="water">
    Thousands of APIs with authentication managed for you.
  </Card>
</CardGroup>

## How authentication works

Aggregators absorb the OAuth complexity. You create an account on the platform, sign in to each service through its interface, and the aggregator issues you a managed endpoint. There are no redirect URIs to configure, no tokens to copy, and no refresh logic to maintain — the credentials stay on the aggregator's side.

<Note>
  Synti still encrypts anything it stores locally, but with an aggregator the sensitive tokens live with the provider. Review what each connected service can access before you rely on it.
</Note>

## Adding an aggregator as a source

An aggregator connects as an MCP server. Add its generated URL to your agent instructions:

```json theme={null}
{
  "mcpServers": {
    "zapier": {
      "url": "https://<your-unique-endpoint>"
    }
  }
}
```

The URL each platform gives you contains an embedded token or connection identifier, so treat it like a secret. Once added, the aggregator's tools are discoverable by the agent immediately.

<Steps>
  <Step title="Create an account">
    Sign up with your chosen aggregator and connect the individual apps you want to reach.
  </Step>

  <Step title="Generate an endpoint">
    Copy the unique MCP server URL the platform issues for your account.
  </Step>

  <Step title="Add it to Synti">
    Paste the URL into an `mcpServers` entry in your instructions, then reload the source.
  </Step>

  <Step title="Test it">
    Ask the agent to list the available tools or run a simple call before building anything complex on top of it.
  </Step>
</Steps>

## What the tools look like

* **Zapier, Composio, and Pipedream** expose pre-built tools scoped to each service you've connected.
* **Make** converts each scenario you've authored into a tool, detecting its inputs and outputs.
* **Composio's universal server** offers a small, fixed set of tools that can reach any app you've linked, rather than one tool per action.

## Things to keep in mind

* Rate limits and available actions differ by platform, so coverage for a given app varies.
* Test a connection with a low-stakes request before wiring it into an automation or a multi-step workflow.
* Because an aggregator adds a network hop and a broad tool surface, keep the agent's permission mode tight for sessions that use one — see [Permissions](/core-concepts/permissions).

## Related

<CardGroup cols={2}>
  <Card title="Sources" icon="database" href="/sources/overview">
    MCP servers, REST APIs, and local filesystems.
  </Card>

  <Card title="Automations" icon="robot" href="/automations/overview">
    Fire agent sessions when events happen.
  </Card>
</CardGroup>