The Model Context Protocol (MCP) is a powerful "universal connector" for AI. It allows you to take the specialized research tools and data from Corbis and use them directly within other AI platforms like Cursor, ChatGPT, Claude Desktop, and Grok.
By connecting Corbis via MCP, you empower your external AI assistants with the same high-quality academic research and economic data that you find within the Corbis platform.
Typically, AI models are limited to the data they were trained on or whatever they can find on the open web. By connecting Corbis via MCP, you provide your AI with "professional superpowers":
When you connect Corbis via MCP, you gain access to 10 specialized tools:
Connecting Corbis to another platform follows a simple three-step process:
Navigate to Settings → API Keys in the Corbis platform. Create a new key (e.g., "Cursor - Laptop") and copy the token immediately. For security, tokens are only shown once.
For ChatGPT: You can use the standard OAuth 2.1 flow. When connecting Corbis as a tool, select "OAuth" and you will be redirected to the Corbis consent screen to authorize the connection.
Different platforms require different setup methods:
https://www.corbis.ai/api/mcp/universalTo protect your data and tools, Corbis requires an API key for all MCP actions. Even if a platform only asks for a URL, ensure your key is included in the "Authorization" or "Bearer Token" field.
Every AI platform handles MCP slightly differently. While Corbis provides the "universal connector," you may need to check the documentation of the specific tool you are using (like Cursor's "Models" settings or Claude Desktop's config file) to ensure the connection is successful.
Your MCP API keys are personal. Never share them or commit them to public code repositories. If you believe a key has been compromised, you can revoke it instantly in your Corbis Settings.
Q: Do I need a paid account to use MCP? A: MCP access is tied to your Corbis account level. Check your entitlements in the settings page to see which tools are available to you.
Q: Why does Cursor need a JSON snippet instead of just a URL?
A: Cursor uses a specific implementation of MCP that requires local command execution (via npx mcp-remote) to establish a secure, bidirectional connection.
Q: Can I use Corbis MCP tools with locally-run AI models? A: Yes! As long as your local AI runner (like Ollama or LM Studio) supports the Model Context Protocol, you can connect it to Corbis.
If you're having trouble connecting Corbis to another platform, reach out to our support team at admin@agenticassets.ai or check the Quick Start Guide for general platform usage.