AI Agents

ExportComments provides first-class support for AI agents through the Model Context Protocol (MCP). AI assistants like Claude, Cursor, and Windsurf can export comments and reviews from 40+ platforms using natural language.

Quick Start

1. Install the package

npm install -g exportcomments-cli

2. Get your API token

Sign up at exportcomments.com and get your token from the API dashboard.

3. Configure your AI client

Add ExportComments as an MCP server. Example for Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "exportcomments": {
      "command": "npx",
      "args": ["-y", "exportcomments-cli"],
      "env": {
        "EXPORTCOMMENTS_API_TOKEN": "your-token-here"
      }
    }
  }
}

4. Start using it

Ask your AI assistant:

"Export comments from this YouTube video: https://www.youtube.com/watch?v=dQw4w9WgXcQ"

The AI will use the MCP tools to create the export, wait for completion, and return the results.

npm Package

The exportcomments-cli npm package provides both the CLI tool and the MCP server:

Requirements: Node.js 18+

Supported AI Clients

See the MCP Server page for detailed setup instructions for each client.

MCP Tools

The MCP server exposes 6 tools that AI agents can call:

See MCP Server — Tool Parameters for the full parameter reference.

Supported Platforms

ExportComments supports 33+ platforms across social media, e-commerce, and review sites:

Social Media: Instagram, YouTube, TikTok, Facebook, Twitter/X, LinkedIn, Reddit, Threads, VK

Video & Messaging: Twitch, Vimeo, Discord

E-Commerce: Amazon, AliExpress, Shopee, Lazada, Flipkart, Etsy, Walmart, Best Buy, eBay

Reviews: Trustpilot, Yelp, Google Reviews, TripAdvisor, IMDb, Airbnb, Steam

Other: Apple App Store, Google Play Store, Disqus, Product Hunt, Change.org

Use the detect_platform tool to check if a URL is supported and see available options, or list_platforms to browse all platforms.

Use Cases for AI Agents

Sentiment Analysis

Export comments from a product page, then analyze sentiment patterns, identify common complaints, and summarize customer feedback.

Competitive Research

Export reviews from competitor products across Amazon, Trustpilot, and app stores, then compare sentiment and feature requests.

Social Media Monitoring

Export comments from social media posts to track brand mentions, engagement patterns, and audience reactions.

Content Research

Export YouTube or Reddit comments to understand audience interests and generate content ideas based on what people discuss.

Market Research

Export reviews from multiple e-commerce platforms to identify market gaps, pricing sentiment, and feature demand.

Example Prompts

Here are example prompts you can use with AI agents that have the ExportComments MCP server configured:

• "Export all comments from this Instagram post and summarize the main themes"
• "Get the reviews for this Amazon product and tell me the top 5 complaints"
• "Export comments from this YouTube video with replies included, then analyze the sentiment"
• "Check what platforms ExportComments supports for this URL"
• "Export the last 100 comments from this Reddit thread and identify the most discussed topics"
• "Get Trustpilot reviews for example.com and create a summary report"

LLM-Friendly Documentation

This documentation is available in machine-readable formats for AI agents and LLMs:

These files are auto-generated from the documentation source and kept in sync with every build. Use llms.txt for a quick overview or llms-full.txt for comprehensive context.

Output Format

All MCP tool responses return structured JSON for easy parsing:

{
  "ok": true,
  "data": { ... }
}

On error:

{
  "ok": false,
  "error": "Human-readable error message",
  "error_code": "MACHINE_READABLE_CODE",
  "detail": "Additional context"
}

Job Lifecycle

When an AI agent creates an export, the job goes through these statuses:

1. queueing — Job is queued for processing
2. progress — Job is actively extracting data
3. done — Job completed, data ready for download
4. error — Job failed (check error details)

Use wait=true on export_comments or check_export to automatically poll until the job reaches a terminal state. The MCP server polls every 5 seconds with a 10-minute timeout.