Mastra AI Agent with Apify MCP Server

The Mastra AI Agent with Apify MCP Server is a generic AI agent designed to demonstrate the power of the mastra.ai framework when paired with the Apify MCP Server. This Actor leverages the Model Context Protocol (MCP) to connect AI-driven workflows with Apify’s extensive library of Actors, enabling seamless data extraction, web scraping, and task automation.

Built with TypeScript and the Apify SDK, this Actor serves as a flexible starting point for developers looking to create custom AI agents that tap into real-time web data and tools.

🎯 Features

• 🤖 AI-powered assistance: Uses the mastra.ai framework to process queries with OpenAI models (e.g., gpt-4o, gpt-4o-mini).
• 🌐 Apify MCP integration: Connects to the Apify MCP Server to access tools like clockworks/free-tiktok-scraper and more.
• 💰 Pay-per-event pricing: Flexible, usage-based billing for Actor runs and task completions.

🔄 How it works

1. 📥 Input

   • Query (e.g., "Analyze posts from @openai and summarize AI trends")
   • Model selection (gpt-4o or gpt-4o-mini)
   • Actors included (e.g., ["clockworks/free-tiktok-scraper"]) - list of Apify Actors to be served by the MCP server outside of the default list

2. 🤖 Processing with Mastra

   • Initializes a Helpful Assistant Agent using the mastra.ai framework.
   • Connects to the Apify MCP Server via Server-Sent Events (SSE).
   • Fetches available tools (e.g., TikTok scrapers, web browsers) dynamically.
   • Executes the query, leveraging MCP tools as needed.

3. 📤 Output

   • Returns the agent’s response as text.
   • Stores results in an Apify dataset with the query and response.

💰 Pricing

This Actor uses the Pay Per Event (PPE) model for cost-effective, usage-based pricing.

Event	Price (USD)
Actor start	$0.10
Task completion	$0.40

Input Example

1{
2  "query": "Analyze the posts of @openai and summarize current trends in AI.",
3  "modelName": "gpt-4o-mini",
4  "actorsIncluded": ["clockworks/free-tiktok-scraper"]
5}

Output Example

Dataset entry:

1{
2  "query": "Analyze the posts of @openai and summarize current trends in AI.",
3  "response": "Here’s a summary of the recent posts from OpenAI's Instagram account, highlighting current trends in AI..."
4}

🔧 Technical highlights

• Mastra.ai framework: Powers the AI agent.
• Apify MCP Server: Exposes Apify Actors as tools via the Model Context Protocol.
• Apify SDK: Ensures seamless integration with the Apify platform.
• TypeScript: Provides type safety and modern JavaScript features.

🤖 Under the hood with mastra.ai

This Actor uses the mastra.ai framework to create a generic AI agent that interacts with the Apify ecosystem:

👤 The agent

• Helpful Assistant Agent:

  1new Agent({
  2  name: 'Helpful Assistant Agent',
  3  instructions: 'You are a helpful assistant who aims to help users with their questions and tasks.',
  4  model: openai(modelName),
  5});
  

  • Goal: Answer queries accurately and leverage MCP tools when needed.
  • Tools: Dynamically fetched from the Apify MCP Server (e.g., clockworks/free-tiktok-scraper).

🔄 Workflow

1. Initialization:

   • Starts the MCP Server if it’s not already running.
   • Connects via SSE using the MastraMCPClient.

2. Execution:

   • Queries the agent with user input.
   • Integrates MCP tools into the agent’s toolset.
   • Processes the response using the selected OpenAI model.

3. Output:

   • Logs token usage and the agent's response.
   • Saves results to the dataset.

🛠️ Tools

Expandable to any Apify Actor from the Apify Store via the MCP Server configuration.

⚙️ Usage

Running on Apify

1. Clone the repository.
2. Push the Actor to the Apify platform:

   apify push
   

3. Set Actor OPENAI_API_KEY environment variable.
4. Run the Actor with your input via the Apify Console or API.

Local development

1. Install dependencies:

   npm install
   

2. Run locally:

   OPENAI_API_KEY=<your-openai-api-key> apify run -i '{"query": "Show me latest post from @openai TikTok profile", "actorsIncluded": ["clockworks/free-tiktok-scraper"]}'
   

📖 Learn more

• Mastra.ai Framework
• Apify MCP Server
• Apify Platform
• Apify SDK Documentation
• What is MCP?
• What are AI agents?
• How to build an AI agent on Apify

🚀 Get started

Ready to build your own AI agent? Deploy this Actor on Apify, tweak the query, and explore the possibilities of mastra.ai and MCP! 🤖✨

🌐 Open source

This Actor is open-source, hosted on GitHub. Want to add features? Open an issue or submit a pull request!