visual-check
Skill0 starsChanged 34 days agoMITView on GitHub
Visually verifies UI changes in a web app. Takes Playwright screenshots of the affected screens, looks at them, and reports JavaScript errors, console errors, failed HTTP responses, blank pages and unexpected redirects. Use when asked for "/visual-check", after finishing a change to anything the user sees (views, styles, components, messages), or to see how a screen looks. Does not produce manuals or videos.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- sergenes/mini_agenttheir own version, first seen Aug 21, 2026
- eriknomitch/workmelttheir own version, first seen Jul 27, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- sergenes/mini_agent★ 29same name
- eriknomitch/workmelt★ 4same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository2 other files it gives its agents.
- user-manualSkill
- video-tutorialSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 19
If $V/node_modules does not exist, run cd $V && npm install and node node_modules/playwright-core/cli.js install chromium.
- Very specific: line 21
Look for docs/webapp-docs.json in the project.
- Very specific: line 41
If the change depends on state (an open panel, an applied filter, a dialog), reproduce it with --click and --type.
- Distinctive: line 66
If something is wrong, fix it, redeploy and take the screenshot again.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Visual check
A UI change is not done until someone looks at the screen. A successful
build, a deploy and an HTTP 200 prove nothing about what the user will see. A
collapsed panel, an icon that fails to load, a half-transparent dialog or
personal data left unblurred only show up in the screenshot.
$V is this skill's scripts/ folder. The user-manual and video-tutorial
skills of webapp-docs-kit reuse these scripts.
Before you start
- If
$V/node_modulesdoes not exist, runcd $V && npm installandnode node_modules/playwright-core/cli.js install chromium. - Look for
docs/webapp-docs.jsonin the project. If it is missing, create it followingreferences/configuration.md. Take the base URL from the project'sREADME.md,CLAUDE.md/AGENTS.mdor run scripts. Ask the user which environment to use if it is not obvious: local, dev or staging. Never use production. - The change must be running in that environment. If it is not, follow the project's own instructions to build and start or deploy it.
Screens behind sign-in or other preconditions
This kit does not sign in, handle credentials or assume user roles. If a screen
you need is behind a login, a paywall, a feature flag, a specific role or any
other precondition (check.js reports a redirect, or the screenshot shows a
sign-in page), stop and ask the user how they want to handle it. It is
their decision. Never ask for, type or store passwords, tokens or cookies
yourself, and never work around access controls.
Steps
- Pick what to check. Take the views touched by the change (
git diff). If the change depends on state (an open panel, an applied filter, a dialog), reproduce it with--clickand--type. Screenshot from the project root:
node $V/check.js <path> [options]Option Purpose --click "SEL1|SEL2"Open panels, tabs, dropdowns and dialogs, in order --type "SEL|text"Type into filters and search boxes --wait MS|SELSlow screens or animated dialogs --fullFull page --mobilePhone viewport (390×844), when the view must be responsive --exists SEL/--text SELCheck an element without looking at the image. Useful to tune selectors --htmlSave the rendered HTML to look for selectors --redact SELBlur more personal data on top of the config's redactlist--mock FILEScreens that need fake API data The screenshot is saved in
~/.cache/webapp-docs-kit/screenshots/<app>/, outside the repository.- Open the image with your file-reading tool. Do not skip this step. Check that the change looks as requested and that nothing around it broke: alignment, clipped text, overlaps, icons, colors and empty states.
- Read the
problems:section of the output. Report JavaScript errors, console errors, HTTP responses ≥ 400, blank pages and redirects. A new error on a screen that looks fine is still a defect. - If something is wrong, fix it, redeploy and take the screenshot again. Compare it with the previous one.
Rules
- Read-only.
--clickand--typeare for filtering, navigating and opening panels.check.jsrefuses to click Save, Delete, Submit, Send, Sign and similar buttons. Use--allow-changeonly when the user explicitly authorizes that specific case, and only in a test environment. --textprints unblurred data to the terminal. Do not copy it into documents or commits.- Known pitfalls (prefixed ids, animated dialogs, charts that render late,
JSF/PrimeFaces, SPAs) are in
references/capture-pitfalls.md.
Report
For each screen checked, give the path, whether it looks right or what is wrong, the problems detected, and the screenshot path. Do not say "verified" unless you opened the image.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

