@aerostack/sdk-vercel-ai

Use Aerostack workspace tools with the Vercel AI SDK.

Aerostack is the full-stack platform for AI agents — compose MCP servers, skills, and functions into a single workspace URL that any AI agent can call. This SDK lets you drop 250+ pre-built tools into your Vercel AI app in 3 lines of code.

Why?

The Vercel AI SDK supports tool calling across any LLM provider — but you still need to build and maintain each tool integration yourself. With Aerostack, you compose a workspace of tools (GitHub, Slack, Notion, Stripe, 250+ more) and this SDK makes them instantly available to generateText, streamText, or any AI provider.

Your App → Vercel AI SDK → @aerostack/sdk-vercel-ai → Aerostack Workspace → GitHub, Slack, Notion, ...

Install

npm install @aerostack/sdk-vercel-ai ai

Quick Start

import { openai } from '@ai-sdk/openai';
import { generateText } from 'ai';
import { getTools } from '@aerostack/sdk-vercel-ai';

const { tools } = await getTools({ workspace: 'my-workspace', token: 'mwt_...' });

const { text } = await generateText({
    model: openai('gpt-4o'),
    tools,
    maxSteps: 5,
    prompt: 'Create a GitHub issue for the login bug',
});

That's it. Tools auto-execute — no handleToolCall needed. The Vercel AI SDK handles the tool-call loop for you with maxSteps.

Streaming

import { streamText } from 'ai';

const { tools } = await getTools({ workspace: 'my-workspace', token: 'mwt_...' });

const result = streamText({
    model: openai('gpt-4o'),
    tools,
    maxSteps: 10,
    prompt: 'Summarize Notion pages and post to Slack',
});

for await (const chunk of result.textStream) {
    process.stdout.write(chunk);
}

Works With Any Provider

import { anthropic } from '@ai-sdk/anthropic';
import { google } from '@ai-sdk/google';

// Claude
const { text } = await generateText({
    model: anthropic('claude-sonnet-4-20250514'),
    tools,
    maxSteps: 5,
    prompt: 'Check GitHub PRs and notify on Slack',
});

// Gemini
const { text } = await generateText({
    model: google('gemini-2.0-flash'),
    tools,
    maxSteps: 5,
    prompt: 'Search Notion and create a report',
});

Factory Pattern

For reusable clients that share a single connection:

import { createAerostackVercelAI } from '@aerostack/sdk-vercel-ai';

const aero = createAerostackVercelAI({ workspace: 'my-workspace', token: 'mwt_...' });
const { tools } = await aero.tools();

API Reference

getTools(config) → Promise<ToolSetResult>

Fetches tools from the workspace and converts to Vercel AI SDK ToolSet format. Each tool includes an execute function that calls the workspace gateway. Returns { tools, raw }.

createAerostackVercelAI(config) → AerostackVercelAIClient

Creates a reusable client that shares a single WorkspaceClient instance.

formatToolResult(result)

Lower-level utility to flatten MCP tool results into strings.

How It Works

1. Tool Discovery — getTools() calls your Aerostack workspace gateway to fetch all connected MCP server tools
2. Format Conversion — MCP tool schemas are wrapped with jsonSchema() from the AI SDK — zero conversion overhead since both use JSON Schema
3. Auto-Execution — Each tool gets a built-in execute function that proxies calls through the workspace gateway to actual MCP servers
4. Error Handling — Errors are caught and returned as string results, keeping the AI conversation flow intact

Requirements

• Node.js 18+
• ai (Vercel AI SDK) >= 3.0.0
• An Aerostack workspace with a token (mwt_...)

Getting Your Workspace Token

1. Sign up at app.aerostack.dev
2. Create a workspace and add MCP servers (GitHub, Slack, Notion, etc.)
3. Copy the workspace token (mwt_...) from the workspace settings

Related Packages

| Package | Framework | |---------|-----------| | @aerostack/sdk-openai | OpenAI SDK | | @aerostack/sdk-langchain | LangChain.js | | @aerostack/core | Core types + WorkspaceClient |

Links

• Aerostack Website
• Documentation
• Dashboard
• MCP Marketplace
• GitHub

License

MIT