You are browsing as a guest. Sign up (or log in) to start making projects!

ContextGuard Lite

By @Siddharth

  • 2 Devlogs
  • 1 Total hours

ContextGuard Lite is a local-first Chrome/Edge browser extension that helps users redact private information before using AI tools like ChatGPT, Claude, and Gemini. It can scan pasted text and simple uploaded files such as .txt, .md, .csv, and .json, then replace sensitive data like emails, phone numbers, passwords, API keys, student IDs, URLs, SSN-like numbers, and card-like patterns with safe labels. Everything runs locally in the browser with no backend, login, paid API, or tracking.

Source
Open comments for this post

@Siddharth on ContextGuard Lite

17m 38s logged

Today I fixed the shipping issue for ContextGuard Lite by creating a proper GitHub Release with a downloadable .crx file.
The project was already working, but the demo link needed to point to an actual release asset instead of only the repository. I packaged the Chrome/Edge extension, uploaded contextguard-lite.crx to the v1.0.0 release, and added a clear release description.
Now testers can download the extension directly from the GitHub Release and try the local-first redaction features, including the popup scanner, file scanner, privacy risk score, and AI-site floating ContextGuard button.
Next, I want to keep improving the inline Grammarly-style underline replacement flow and test it on more AI websites.

0
0
8
Ship #1

@Siddharth on ContextGuard Lite

I made ContextGuard Lite, a local-first browser extension that helps people remove private information before pasting text into AI tools.

The extension detects things like emails, phone numbers, API keys, passwords, URLs, student IDs, SSN-like numbers, and card-like patterns. It can replace them with safe labels like [EMAIL] or [PHONE]. I also added a Grammarly-style feature where risky text can be underlined, then clicked or hovered over to replace it.

The hardest part was making the detection accurate without sending anything to a server. I’m proud that it has no backend, no login, no tracking, no paid API, and no API keys.

To test it, load the extension in Chrome/Edge, open the popup, paste text with private info, and click the redaction button. You can also test the underline replacement flow on supported AI text boxes.

  • 1 devlog
  • 0h
  • 3.47x multiplier
  • 1 Stardust
Try project → See source code →
Open comments for this post

@Siddharth on ContextGuard Lite

19m 9s logged

Today I worked on making ContextGuard Lite more polished and closer to production-ready.
I reviewed the project files, checked the extension flow, and improved the idea for the Grammarly-style privacy warnings. The goal is for detected private information to be underlined directly in the page or text area, then show a small popup when the user hovers or clicks it. From there, the user can replace the risky text with a safe label like [EMAIL], [PHONE], [API_KEY], or [PASSWORD].
I also focused on keeping the extension local-first. There is still no backend, no login, no tracking, and no paid API. Everything should run inside the browser so users can redact private information before pasting text into AI tools.
Next, I want to test more websites, improve false-positive detection, and polish the popup/tooltip UI.

0
0
23

Delete project?

Are you sure you want to permanently delete this project? This action cannot be undone.

All devlogs, followers, and associated data will be removed.

Followers

Loading…