liquid-glass-suite
AGENTS.md0 starsChanged 7 days agoMITView on GitHub
This repository is Liquid Glass Suite — Apple Liquid Glass agent skills for web pages, packaged as a Claude Code plugin and an Agent Skills collection. Do not copy a single skills/<name> folder as a standalone skill. All four skills reference ../../shared/..., so the suite must stay together. Generic Agent Skills CLI: That installs the repository as one skill carrying the whole suite, because the CLI copies only a skill folder and the four task skills reference ../../shared/.... The installed…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 33,315 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository6 other files it gives its agents.
- docs/llms.txtllms.txt
- liquid-glass-suiteSkill
- liquid-glass-debugSkill
- liquid-glass-nativeSkill
- liquid-glass-opticsSkill
- liquid-glassSkill
The file
We split this AGENTS.md into the rules it gives an agent
5 rules. About 0 are common, 0 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Do not copy a single skills/<name> folder as a standalone skill.
- Very specific: line 38
Run npm ci && npm run verify before claiming a pass.
- Very specific: line 39
Read shared/references/15-honesty-and-checklists.md before quoting any numeric claim.
- Very specific: line 45
Do not add a fifth folder under skills/ without updating SKILLS in tests/test_package.py.
- Very specific: line 47
Do not add a skills key to .claude-plugin/plugin.json.
≈ “Common” is an estimate: at least 1 in 5 of the 42,283 AGENTS.md 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
AGENTS.md
This repository is Liquid Glass Suite — Apple Liquid Glass agent skills for web pages, packaged as a Claude Code plugin and an Agent Skills collection.
Install it intact
Do not copy a single skills/<name> folder as a standalone skill. All four
skills reference ../../shared/..., so the suite must stay together.
/plugin marketplace add admos98/liquid-glass-suite
/plugin install liquid-glass-suite@admos98-skillsGeneric Agent Skills CLI:
npx skills add admos98/liquid-glass-suiteThat installs the repository as one skill carrying the whole suite, because the
CLI copies only a skill folder and the four task skills reference ../../shared/....
The installed entry (liquid-glass-suite) is a router: do the task from
skills/<name>/SKILL.md inside the installation.
The four skills
| Skill | Job |
liquid-glass | Apply: drop-in CSS/JS recipes for nav, tabs, cards, sheets, menus, buttons, search, sidebar, sliders |
liquid-glass-optics | CSS/SVG displacement, reflective rim, WebGL2, the standalone showcase |
liquid-glass-debug | S01–S33 symptom triage, Chromium/WebKit quirks, pixel measurement |
liquid-glass-native | SwiftUI/UIKit/AppKit, DocC-checked declarations, availability fallbacks |
Rules for working in this repository
- Run
npm ci && npm run verifybefore claiming a pass. Report what actually ran. - Read
shared/references/15-honesty-and-checklists.mdbefore quoting any numeric claim. Apple publishes no numeric Liquid Glass specification. - Native Apple samples are not compiled here. Do not report a native-runtime pass.
- Playwright Chromium is not a real iPhone. Do not infer device behaviour from it.
tests/test_package.pyenforces contracts, including byte budgets onREADME.mdand eachskills/*/SKILL.md. Re-run the suite after editing either.- Do not add a fifth folder under
skills/without updatingSKILLSintests/test_package.py; the test asserts exactly four folder names. - Do not add a
skillskey to.claude-plugin/plugin.json; standardskills/discovery is asserted.
Layout
| Path | Contents |
skills/ | The four published skills |
shared/assets/ | CSS/JS drop-in, WebGL2 renderer, standalone showcase |
shared/references/ | Sixteen implementation, browser, measurement and honesty references |
shared/scripts/ | Displacement-map, edge-profile and chroma-retention tools |
tests/ | Package contracts, browser suite, evidence transcript |
evals/ | Ten agent benchmark prompts (results table intentionally empty) |
docs/ | README imagery, comparison, and the Pages landing files |
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.

