RAI - AI Interaction Firewall

RAI is a browser extension that scans your AI conversations in real time for security threats. It runs entirely on your device with zero network calls. What it detects: - Prompt injection attacks (jailbreaks, role overrides, instruction manipulation) - Credential and secret exposure (API keys, tokens, passwords in AI responses) - Infrastructure threats (supply chain indicators, model poisoning signals) Supported platforms: - Claude (claude.ai) - ChatGPT (chatgpt.com) - Google Gemini (gemini.google.com) Privacy first: - Zero data leaves your device. All scanning uses local pattern matching. - No analytics, no telemetry, no external API calls. - Open source: https://github.com/pektimole/rai Strict Mode: Enable strict mode to block paste operations when critical threats are detected, preventing you from accidentally sending compromised prompts.