render-check
Skill2 starsChanged 3 months agoMITView on GitHub
Verify a frontend change ACTUALLY rendered before calling it done. Inspect the served HTML + compiled CSS on the live dev server — never trust source edits or a normal browser refresh. Catches the dev-server staleness trap (Fast Refresh silently drops edits) that makes every "fix" look identical to the last. Use after any UI/layout/Tailwind change, and as the gate before saying "done".
- Deletes or force-pushes
1 other project uses a version of this.
Adopted by
Most recently taken up by
- kiloloop/kiloloop-skillstheir own version, first seen Aug 15, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- kiloloop/kiloloop-skills★ 0same 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.
- agent-designSkill
- dual-tool-handoffSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 28
Do not hardcode ports or paths.
- Distinctive: line 44
If the change touched layout/Tailwind and the result looks unchanged, assume staleness first — restart clean rather than re-editing CSS:
- Distinctive: line 110
Only say a visual change is done when you can state, concretely:
- Distinctive: line 116
If you can't produce those two grep results, the honest status is NOT done — unverified, not "done".
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
Render Check
The source file is not the truth. The served markup + compiled CSS are the truth.
Why this skill exists
AI agents love to say "done" after editing a React/Tailwind file. Sometimes the source changed but the browser is still serving a stale build (Next.js Fast Refresh quietly stopped updating). Typecheck passes. Curl returns 200. The screen doesn't change.
Iron rule: do not tell the user a visual change is done until the served HTML contains the new markup AND the served CSS contains the new rule. "I edited the file and tsc passes" is not evidence the change rendered.
Discover project specifics first
Do not hardcode ports or paths. Before checking:
- Find how this repo starts the dev server (package.json scripts, README, CLAUDE.md).
- Note the port and the route under test.
- Note the build-cache dir (often
.next/for Next.js;.vite//dist/for Vite). - Find the served CSS link from the HTML — don't invent the path.
Protocol (run before confirming a frontend change)
1. Make sure the server is alive and freshly compiled
# Replace PORT with the project's dev port
lsof -ti :PORT >/dev/null 2>&1 && echo "up" || echo "down"If the change touched layout/Tailwind and the result looks unchanged, assume staleness first — restart clean rather than re-editing CSS:
lsof -ti :PORT | xargs kill -9 2>/dev/null
rm -rf .next # or this project's cache dir
# restart via the project's usual dev command (background)
until curl -fsS -o /dev/null "http://localhost:PORT/<route>"; do sleep 2; done2. Confirm the new MARKUP is in the served HTML
Grep the live page for the exact classes/attributes/text you just added:
URL="http://localhost:PORT/<route>"
curl -sS "$URL" --max-time 30 | grep -c "<the new class or text>"0 = the change did not render. Stop. Restart the server (step 1). Re-check. Do NOT keep
editing source — the source may already be correct.
Grep gotcha — interpolation markers: React may insert <!-- --> around {interpolations},
so a substring that straddles a dynamic value won't match raw HTML. Grep a purely-static
substring, or strip tags first:
curl -sS "$URL" | python3 -c "import sys,re,html;print(html.unescape(re.sub(r'<[^>]+>',' ',sys.stdin.read())))" | grep -c "<text>"Grep gotcha — RSC / hydration payloads: App Router and similar frameworks embed data as JSON in the HTML. A string can appear in the count even when it's not in the visible DOM. To verify what actually renders, inspect the tag-stripped text region around an anchor — don't trust total counts alone.
3. Confirm the new CSS RULE was generated
A class in the HTML does nothing if the CSS toolchain never emitted the rule (Tailwind JIT only generates classes it sees at build time):
# Next.js example — adapt the CSS URL pattern to this stack
CSS=$(curl -sS "$URL" --max-time 30 | grep -oE '/[^"]+\.css' | head -1)
BODY=$(curl -sS "http://localhost:PORT$CSS" --max-time 30)
echo "$BODY" | grep -c '<utility-name>' # expect >= 10 = CSS never picked up the class → stale build or an unscanned file. Restart clean (step 1).
4. (When layout is the doubt) verify the COMPUTED result
Markup + rules present but it still looks wrong = a real CSS logic bug (not staleness). Only now reason about the CSS. If a headless browser is available, screenshot at a desktop width and inspect computed styles. Common height-chain culprits:
- A flex/grid child needs
min-h-0to scroll instead of growing. - A grid filling a height needs its row set to
1fr, not auto. h-full(percentage) is flaky across flex nesting — prefer stretch.
5. Tell the user to hard-refresh
Even with the server fixed, their tab caches old CSS/HTML. End with: "hard-refresh (Cmd+Shift+R / Ctrl+Shift+R)" — a normal refresh often won't pull the new asset.
Done gate
Only say a visual change is done when you can state, concretely:
- served HTML contains
<new markup>(grep count ≥ 1), and - served CSS contains
<new rule>(grep count ≥ 1), and - the user has been told to hard-refresh.
If you can't produce those two grep results, the honest status is NOT done — unverified, not "done".
Output format
Report the evidence, not the intent:
RENDER CHECK — /<route>
server: restarted clean / already fresh
markup: <token> ×N ✓/✗
css: <rule> present ✓/✗
computed: inspected / not inspected
action: hard-refresh your tab
verdict: RENDERED ✓ / NOT RENDERED — restarted, re-checkDoes 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.

