Inspiration

Imagine you are studying for an exam and you have a specific concept in mind that you want to study. Unfortunately, you can't remember what lecture or textbook chapter it was covered in. Your first thought might be to open each lecture slide deck in a new tab and press Ctrl+F on each one.

If you've been there before, you know Ctrl+F doesn't help much. A one or two word query returns hundreds of matches; and a long, specific one returns none. It also only searches the page you're on, and information is often nestled down a rabbit hole of links to other pages. We wanted to take the burden of navigating a site's structure off the user, so that finding one relevant piece of information doesn't require understanding the whole site.

That's why we built CtrlF++.

What it does

CtrlF++ is a browser extension that makes searching for text easy. Just press CTRL+Shift+F and type what you're looking for. Instead of searching for exact text matches on the current page, it uses vectored search to find text that is semantically similar to your query. It also uses the page you're already viewing as context to dig deeper and find relevant links, follow them, and surface the exact passages you need on other pages!

It can:

  • Discover relevant resources: Identify links and connected pages worth exploring based on the current page's context.
  • Explore recursively: Follow relevant connections to discover information buried deeper within a website.
  • Search inside documents: Find relevant content in supported PDF, DOCX, and PPTX files
  • Find information, not just text: Use exact and semantic matching to locate relevant passages, even if they share no words with the original query
  • Show discovery provenance: Highlight matching content and show its original source and the path taken to discover it
  • Keep users in control: Provide a read-only discovery experience that doesn't autonomously submit forms or take actions on websites.

Our goal is simple: less navigating, more finding!

Privacy by design

We believe making information easier to discover shouldn't cost you your privacy.

CtrlF++ processes extracted page and document content locally, inside the extension. Semantic matching uses an on-device model rather than sending extracted content to external LLMs or AI services.

CtrlF++ is also read-only by design; it discovers and highlights information without autonomously interacting with websites on the user's behalf.

How we built it

We built CtrlF++ as a browser-native extension with a local-first processing architecture.

The discovery engine starts from the active page, examines its context and links, and explores relevant connected resources. It processes supported documents inside the extension and uses local semantic matching to identify relevant information.

The discovery interface presents matches alongside their source and exploration path, helping users understand not only what was found, but where it came from.

Challenges we ran into

One of our biggest challenges was making recursive discovery useful rather than simply following every link on a website. Exploring too many unrelated resources would defeat the purpose of a quicker search. We needed to make discovery context-guided, prioritize relevant connections, and keep exploration within defined boundaries.

Another challenge was supporting different resource types. Information doesn't live only in HTML pages; it can be buried inside PDFs, Word documents, and PowerPoint presentations. Making these resources searchable required handling different extraction and processing paths.

What's next for CtrlF++

We want to make CtrlF++ useful across more information-heavy websites and resource types, improve discovery relevance, and continue strengthening the transparency and privacy guarantees of its local-first architecture.

Don't just search the page, discover the information behind it.

Built With

Share this project:

Updates

Submission history