Redact
Automatically detects and obfuscates sensitive data in text and HTTP requests, preventing exposure to AI models during analysis.
Automatically detects and obfuscates sensitive data in text and HTTP requests, preventing exposure to AI models during analysis.
Redact is an MCP server and Claude Code plugin designed for penetration testers, enabling them to leverage AI analysis without compromising client data. It employs a dual-detection approach, combining comprehensive regex patterns and AI-powered Named Entity Recognition (NER), to identify and replace sensitive information like IPs, hostnames, emails, API keys, personal names, and organization names with consistent, deterministic fake values. The server maintains a bidirectional mapping table, ensuring that real values can be restored for client reporting, and offers an audit log for compliance. It seamlessly integrates into Claude Code workflows, proxying requests and responses to ensure no real PII is exposed to the AI.