Optional Capture · WXT beta
A page you can see.
A record you can use.
Capture a page in your own signed-in browser session. Focus on one region, review the screenshot, then turn declared readings into structured data with a browse course.
Choose where the browser runs
Hosted sandbox
Run scheduled cams without an extension. The service opens an isolated browser and handles its vault connection and saved session. Hosting and model usage determine the running cost.
Optional browser extension
Capture manually from a browser you already signed into. Capture runs on your device; the extension does not request passwords, vault tokens, or cookies.
Use local CaptureBoth paths produce evidence for the same courses. Hosted infrastructure estimates are available; a customer billing flow is not yet connected.
One reader. Two extension builds.
WXT packages the shared browse.cam reader for Chrome and Safari. Version 0.5.1 opens a toolbar popup with teaching, source watching and local English OCR. No browse.cam or dots login is needed. The side panel is optional when the browser supports it.
Chrome desktop
Unpacked beta for Chrome 116+. Ready to install for a signed-in capture test.
Download Chrome build ↓- Unzip, open
chrome://extensions, enable Developer mode, then select Load unpacked. - Open your signed-in page and click the browse.cam toolbar icon to open its popup.
- Choose Teach this page to demonstrate a course, or choose a source to watch its text and read a focused capture.
Updating an unpacked install? Download and unzip the new build, replace the contents of the folder you originally loaded, then click Reload on its extension card. Other Chromium browsers need browser-specific testing; Firefox builds are not yet distributed.
Safari · Mac, iPhone and iPad
Shared popup build, targeting Safari 17+. This ZIP is web-extension source for Apple packaging. Device installation and Safari acceptance tests are still required.
Download Safari build ↓Chrome on Android or iPhone: ordinary mobile Chrome does not run desktop extensions. Use hosted cams through the web app.
Both downloads include offline model assets, about 15 MB each. OCR starts on demand and reports initialization and reading time. Captures stay local until you save or share them. Moondream and other vision language models are not bundled in this release.
Try on-page course teaching →Create a course for repeat readings →Load its JSON in the extension before capturing. The local reader checks the selected route, account markers, freshness and field types.
browse.cam for agent-browser
Record selected captures and actions, recall their structured data, and propose courses from repeated runs. Frames stay private and are deleted after processing, with scheduled cleanup for interrupted runs. Your readings remain in your own memory.
Local plugin beta. Connect from Settings after server storage is configured. Course candidates require review; this plugin does not perform page actions.
Daily courses with persistent sessions
The proposed workflow wakes a hosted session, reads current data, performs a permitted action, and verifies a receipt. Scheduled captures exist today; automatic claims, submissions and action receipts are still being designed.
Estimate a course’s running costReview a capture
Choose a saved capture JSON. Your file stays in this browser.
page → focused frame → typed fields
For a match: teams, score, match status, goal events, and team stats. For an account: its identity and the specific values you need.
This beta captures on request. Scheduled cams use a separate cloud browser session.