Screens, manuals and evidence
Pictures that people use: a page at every screen size, a user guide written from a task you did once, a GIF for a help article, scores that show whether a page got better, and a history of how a watched page changed.
Screens: every screen size at once
kin-browser sweep https://your-app.com --devices all # 11 sizes, problems named
kin-browser sweep https://your-app.com --theme both --full # light and dark, whole pages
kin-browser screens --current # your tab as it is now, every size, PNGOr the Grid → Screens: pick sizes, press Take the pictures, zoom in, save one or all.
| Option | What it does |
|---|---|
--devices | phone-small 375×667, phone 390×844, phone-large 430×932, android 412×915, tablet 820×1180, tablet-landscape, ipad-pro 1024×1366, laptop 1366×768, laptop-hd 1440×900, desktop 1920×1080, wide 2560×1440, or all |
--sizes 1280x800,… | Your own sizes |
--landscape | Phones and tablets turned sideways |
--theme light|dark|both | The page's light and dark looks |
--full | The whole page, not just the first screen |
--png | Sharp pictures for a manual (JPEG otherwise) |
--scale 2 | Retina-sharp (2×) |
--emulate | Each size as that device's own browser: its identity, touch, pixel density |
--browsers chromium,firefox,webkit | Firefox's and Safari's engines too (kin-browser install-browsers firefox webkit first) |
--country SG | As seen from a country, through a proxy a person set (kin-browser proxies add SG …) |
--current | Your tab as it is — a menu open, signed in, a step into a form — without reloading |
Every run names what breaks, by element: anything wider than the screen
(and which element sticks out, by how far), text under 12px, and tap
targets under 24px on phones and tablets. Each run saves an index.html
contact sheet of all the pictures.
Marked-up pictures
kin-browser annotate e12 e15 --hide-personal --out step.pngBoxes and numbered badges on the elements you name; --mask hides elements;
--hide-personal hides every email, phone number and ID number on screen.
User manuals from a task
- Turn recording on (Grid → Record all agent work, or for one space).
- Do the task — or let an agent do it.
- Recordings → the recording → Make the guide, or:
kin-browser manual last --title "How to submit an expense claim" --languages en,ms,zh
kin-browser manual --recipe check-invoice --phoneKin Browser replays the steps in a separate browser with your sign-ins, takes a sharp picture of each screen before each step, boxes and numbers the thing to press, and writes the guide in English, Bahasa Melayu and 中文: a web page, Markdown (for a wiki or help centre) and a PDF. A step it must not do — typing a password — is described in words ("type your own password"), and the guide carries on.
GIFs and videos
kin-browser export-video last # a GIF, each step's words under its picture
kin-browser export-video last --webm # a videoOr Recordings → Save as GIF / Save as video.
Scores
kin-browser score https://your-site.comPerformance, accessibility, SEO and best practices, each out of 100, with what to fix first. Kept per page, so the next run says what moved: "SEO 70 (+15 since 18 Oct)".
Page history
Grid → History: each run of a watched or scheduled page, with the picture it took. The newest two are compared side by side straight away, with the lines added and removed; pick any two to compare.
Tests from a crawl
kin-browser crawl http://localhost:3000 --write-tests kin-tests/
kin-browser test kin-tests/A smoke test for every page that answered: it opens, shows its title, and (when it had none) no script errors. Add your own checks to it.
Point and pick
In Kin Browser — or Chrome with Kin Browser for Chrome — select text on any page, right-click → Kin Browser:
- Tell me if this disappears / when this appears again — a watch, hourly
- Add a test check for this text — into
tests/picked.kin.yaml - Save as a note about this site — what agents are told when they open it