Chrome extension that blocks API keys from being pasted into AI tools
Local regex scanning blocks secrets before they hit ChatGPT, unlike cloud-based DLP tools.
Chrome extension that masks secrets & sensitive data before pasting into AI chats. Works with ChatGPT, Claude, Gemini & more. 100% local, privacy-first.
Intercepts pastes, masks 30+ secret types locally—zero network requests or tracking.
Developers debugging code in AI chats
1Password's watchtower alerts · GitGuardian · detect-secrets CLI tool
So I built (with some help from Claude) a simple Chrome extension that intercepts the paste, detects secrets using local regex, and replaces them with [MASKED] before they reach the chat. Originals stay in a local AES-256 encrypted vault for unmasking.
No servers. No network requests. No tracking. ~41 KB, zero dependencies. Don't take my word for it: 'grep -r "fetch\|XMLHttpRequest" content_script.js' returns nothing.
Works on ChatGPT, Claude, Gemini, Grok, Perplexity, DeepSeek, and any custom site you add. Supports 30+ patterns — AWS keys, GitHub tokens, JWTs, Stripe keys, database URLs, private keys, and more. You can toggle individual patterns off for false positives.
Open source, MIT licensed. With the recent news about extensions harvesting AI conversations, I figured more devs could use this.
Would love feedback — especially on patterns I might be missing or edge cases you hit.
Local regex scanning blocks secrets before they hit ChatGPT, unlike cloud-based DLP tools.
eBPF network interception beats sidecars for zero-code secret injection.
Four-pass detection with deterministic fakes beats simple regex redaction tools.
MITM proxy catches leaked secrets before they hit AI APIs — better than post-hoc scanning.
Drop-in Slack alternative with verifiable crypto, but one-time secret sharing exists.
Local secret redaction for AI IDEs when cloud-based scanners can't intercept prompts.