Hi Product Hunt! I built Mamorukun (まもるくん) to add a practical check at the moment sensitive information is about to enter an AI chat. It checks text locally in your browser, explains detected items, and lets you select what to replace with placeholders before sending to ChatGPT, Gemini or Microsoft Copilot. You remain responsible for reviewing anything it does not detect or you choose not to mask. Attachments work differently: Mamorukun inspects the text it can extract from supported files and reports unsupported or uninspected content. It does not redact or rewrite the file. If you explicitly continue, the original file goes to the AI service. The extension is free, needs no account or API key, and currently focuses on Japanese workflows. Detection and masking run locally; your chosen AI service still receives the text or file you decide to send. Mamorukun existed before the Astra Challenge. GPT-6 Astra in Codex helped review detection and file-handling boundaries, prioritize fixes, check the implementation and verify the website's explanations. Development also involved Gemini in Antigravity. I'd value feedback on detection accuracy and whether the send-time confirmation is clear enough for everyday work.

Mamorukun

Mask sensitive details before sending them to AI

Mask sensitive details before sending them to AI