Glint

Glint catches sensitive data in your prompts before it reaches AI chat tools. Developers, analysts, and knowledge workers paste secrets into LLMs every day — sometimes deliberately (debugging a stack trace), sometimes by accident (a config file with embedded credentials). Once that data hits a third-party AI server, you cannot pull it back. Glint sits between you and the send button. HOW IT WORKS 1. You type or paste into a supported AI chat field. 2. A small shield icon appears next to the field. Green means clean. Red means sensitive data detected. 3. If you try to send a prompt containing sensitive data, Glint intercepts the send action and shows a modal listing what it found. 4. You choose: Send Anyway (override) or Replace and Send (Glint redacts the secrets with placeholders and sends the cleaned version). WHAT GLINT DETECTS Glint 35+ sensitive data patterns: - Cloud credentials: AWS access keys, GCP API keys, payment provider keys - AI provider API keys - Developer tokens: source control PATs, package registry tokens, team chat tokens - Authentication: JWTs, bearer tokens, Authorization headers, OAuth client secrets - Structural secrets: PEM private key blocks, URLs with embedded credentials, database connection strings - Labeled values: anything written as password: ..., api_key = ..., secret is ... - Personal data: email addresses, US-format SSNs, card-like number sequences SUPPORTED PLATFORMS Glint works on all major AI chat interfaces. The complete list of supported domains is published in the open-source repository. PRIVACY Glint runs entirely in your browser. No data is collected, transmitted, or stored. No analytics. No accounts. No server exists. The source code is open and publicly auditable. Verify these claims yourself. LIMITATIONS Glint uses pattern matching, not machine learning. It catches things that look like secrets. It will occasionally miss confidential prose or proprietary business logic, and may occasionally flag strings that resemble credentials but are not. Use Glint as a safety net, not a guarantee. If you handle regulated data, pair it with organizational policy. OPEN SOURCE MIT licensed. Source code, issue tracker, and roadmap available publicly. BUILT BY Digits Digital — Harare, Zimbabwe.