> SpecStream is the live web data engine for teams and agents. This is the Markdown copy of https://specstream.ai/docs/mcp/make. Every page, with what SpecStream does and costs: https://specstream.ai/llms.txt

# Use SpecStream in Make AI Agents

Add SpecStream to an agent in Make AI Agents as an MCP server. The agent asks for live web data in your words, keeps to the price limit you give it, and hands the results to the next app in your scenario, such as Google Sheets.

## What you need

-   A SpecStream account. [Start free](https://specstream.ai/signup): $10 of work for 30 days, no card.

The server address is the same for every client:

Server address

```text
https://api.specstream.ai/mcp
```

## Connect

### Create an API key

1.  Open [Agents & API](https://specstream.ai/workspace/agents?tab=keys) in the workspace and choose **API keys**.
2.  Choose **Create API key**. Give it only what the task needs: read, prepare or execute permissions, the projects it may use, an expiry and a spending limit.
3.  Copy the key when it's shown. It isn't shown again.

### Add the server to your agent

1.  In the Scenario Builder, add the **Make AI Agents (New)** app's **Run an agent** module, or open a scenario that has one.
2.  Hover over the plus icon on the module and click **Add MCP**.
3.  In **Connection**, click **Add**. Under **MCP Server**, choose **\+ New MCP Server**, paste the server address in **URL** and your key in **API Key / Access token**, then save.
4.  In **Tools**, choose the SpecStream tools the agent may use and click **Save**. For a request that ends in rows, `compile`, `quote_capacity_work`, `authorize_capacity_work`, `get_find` and `list_find_rows` are enough. Make counts every tool you add against the agent's AI tokens.

### Add the results to Google Sheets

1.  On the same **Run an agent** module, click **Add tool**, search for Google Sheets, and choose **Add a Row**.
2.  Connect your Google account, pick the spreadsheet and the sheet, and save. Give the sheet a header row with the columns you ask for.
3.  Give the agent the request below. It gets the quote first and approves it only within the maximum you name, and the key's spending limit holds whatever the agent does.

To run the same request on a schedule without an agent, use the [SpecStream app for Make](https://specstream.ai/integrations/make.md). It sends one bundle per record, marked new or changed.

## Try it

Once Make AI Agents shows SpecStream's tools, ask in your own words:

> Use SpecStream to find every roofer in Columbus, Ohio rated 4.5 or better, with their phone and website. Approve the quote only if its maximum is $5 or less. Then add each result as a row in the Google Sheet Roofers, with its name, phone, website and source URL.

SpecStream reads the request back and gives you the price. Nothing is spent until you approve a maximum, and you pay per match, each with its source.

## Set access and spending limits

An agent works on your account, with the same balance, permissions and spending limits you have. Give its key the least it needs: read, prepare or execute permissions, project limits, an expiry and a spending ceiling. Paid work needs an approved price with a ceiling, and reading, exporting and sharing cost nothing.

## More

-   [MCP connection details](https://specstream.ai/docs/mcp.md), with every tool your agent can call.
-   [The REST API](https://specstream.ai/docs/api.md), [the SDKs and command line](https://specstream.ai/docs/sdk.md), and [everything your agent can do](https://specstream.ai/agents.md).
-   [How quotes and billing work](https://specstream.ai/docs/billing.md).
-   Other clients: [Claude](https://specstream.ai/docs/mcp/claude.md), [ChatGPT](https://specstream.ai/docs/mcp/chatgpt.md), [Cursor](https://specstream.ai/docs/mcp/cursor.md), [VS Code](https://specstream.ai/docs/mcp/vscode.md), [Perplexity](https://specstream.ai/docs/mcp/perplexity.md), [Gemini](https://specstream.ai/docs/mcp/gemini.md), [Grok](https://specstream.ai/docs/mcp/grok.md), [Le Chat](https://specstream.ai/docs/mcp/le-chat.md), [Manus](https://specstream.ai/docs/mcp/manus.md), [LM Studio](https://specstream.ai/docs/mcp/lm-studio.md), [Cherry Studio](https://specstream.ai/docs/mcp/cherry-studio.md), [AnythingLLM](https://specstream.ai/docs/mcp/anythingllm.md).

[Get started – free](https://specstream.ai/signup)

Make AI Agents's own documentation: [Make: Create your first AI agent](https://help.make.com/create-your-first-ai-agent), [Make: MCP Client](https://apps.make.com/mcp-client).