If you discover a security vulnerability related to Claude.ai or MCP integration, please report it through Anthropic's responsible disclosure program:

https://www.anthropic.com/responsible-disclosure

Please do not report security vulnerabilities through public GitHub issues.

This repository is for announcements and community feedback about MCP integration in Claude.ai. Security issues that should be reported include:

• Vulnerabilities in Claude.ai's MCP Connector
• Authentication/authorization bypass in MCP integrations
• Data exposure through MCP tools or resources

• Issues with the MCP protocol specification (report to modelcontextprotocol/specification)
• Issues with third-party MCP servers (report to the server maintainer)
• General Claude.ai security issues unrelated to MCP