8 October 2026 — Plus Plan Available
- AI Prompt Leak Guard Plus is available for INR 99 per person per month, including any applicable tax, one person per verified buyer account.
- Buy through checkout with a verified buyer email and Razorpay. Redeem the emailed activation on the activation page and paste the signed entitlement into Options → Plan. Each monthly renewal emails another activation that must be imported manually.
- Plus adds local inspection of readable selected uploads. Free local prompt protection stays free.
Unreleased
- New detectors: GitLab access tokens, AWS secret access keys and Azure SAS tokens.
- Custom sensitive terms: add your own codenames or client names in Settings. They are matched locally, stored as plain text on your device and blocked by default.
- Hidden characters (zero-width, soft hyphen, fullwidth forms) are normalized before detection.
- Faster checks on very long prompts, more international phone numbers, and fewer false positives for 9-digit numbers and words like "sow".
- Fixed a build issue that could stop the page script from loading in Chrome.
These changes are not yet in a store-submitted build.
0.1.0 — 7 October 2026 — Release Candidate
- Local prompt warnings, redaction and blocking on supported browser AI pages.
- Popup, settings and side-panel views with light, dark and system appearance.
- Local site switches, detector settings and hashed trusted-value entries.
- Stricter policy imports and upload-review handling for unreadable content.
- Separate sandbox activation with signed entitlements, email recovery and codes valid for up to 24 hours.
- Corrected descriptions of browser site access, local storage and the boundary between free protection and sandbox billing.
The standard candidate remains free local mode. Sandbox billing uses test payments and cannot activate the standard candidate. Public paid sales and browser-store publication are pending.
Known Limits
Site layouts change and detectors can miss sensitive values or produce false positives. Review prompts yourself. Desktop/mobile apps, cloud team sync and managed-policy publishing are not provided. Plus upload inspection can inspect readable text; compressed, incomplete, image or otherwise unreadable content requires review. It does not certify files as safe.
Earlier Development
Earlier September and April builds introduced the local protection UI, site adapters and evaluation controls. Historical screenshots and automated fixture results describe those builds and are not proof that every current live site has been verified.