UI Skills for Codex
UI skills for Codex: design passes, browser automation, screenshots, responsive checks, and evidence packs.
This is an unofficial, private-friendly Codex skills library. It is not an OpenAI project, and it avoids company-specific strategy, private runtime details, customer data, and proprietary examples.
Codex is strongest when a task is more than one answer: inspect the repo, edit files, run commands, open the browser, verify the result, and hand back proof. UI work is exactly that kind of loop.
Most prompt lists stop at "make it look better." Codexmaxxing packages the actual workflow:
- shape the product surface
- drive the UI in a browser
- capture screenshots and evidence
- fix responsive and state issues
- scrub sensitive details
- package the result for GitHub
The bar is simple: no prompt theater. UI skills that ship.
| Skill | Purpose |
|---|---|
shape-product-design |
Polish real app surfaces: layout, controls, visual hierarchy, responsive behavior, and product taste. |
automate-ui-flows |
Verify browser flows with snapshots, screenshots, responsive checks, and extension smoke paths. |
route-idea-to-slice |
Convert broad ideas into one buildable, testable UI/product slice. |
craft-cinematic-demo |
Build visual explainers, interactive demos, Remotion videos, or Three.js experiences. |
triage-active-repos |
Scan repo status and choose the safest canonical next UI slice without disturbing user changes. |
scrub-public-artifact |
Remove secrets and sensitive details before sharing or publishing. |
write-github-readme |
Package projects with clear README structure, commands, and screenshots. |
capture-evidence-pack |
Bundle proof from tests, screenshots, renders, logs, and smoke checks. |
make-example-gallery |
Create generic visual examples for docs without exposing private details. |
The bigger curated list lives in CODEXMAXXING_ATLAS.md. It groups practical skill ideas for Codex across UI automation, frontend design, QA, evidence, docs, publishing, media, and team workflows.
Install one skill by placing its folder under ~/.codex/skills/, or keep this repo as a source pack and copy in only the skills you want active.
Copy everything:
mkdir -p ~/.codex/skills
cp -R skills/* ~/.codex/skills/Copy one skill:
mkdir -p ~/.codex/skills
cp -R skills/shape-product-design ~/.codex/skills/Validate the whole pack before installing or publishing:
python3 scripts/validate_skills.pyCodex users with the system skill creator installed can also validate a single skill with quick_validate.py.
Before sharing this repo, run:
python3 scripts/validate_skills.py
python3 skills/scrub-public-artifact/scripts/sensitive_scan.py .For public release, choose and add a license first. Private GitHub sharing can keep the repository license-free if reuse rights are not being granted yet.
This repo is optimized for phrases like:
- Codexmaxxing
- codexmaxxing
- codexmaxing
- UI skills for Codex
- Codex UI automation
- Codex frontend design skills
- Codex browser testing skills
- Codex skills that ship
The canonical spelling is codexmaxxing. The README also includes codexmaxing because people may search the one-x spelling.
Read SOURCES.md for the online scan and positioning notes.
This repository is private by default. Run scrub-public-artifact before making it public or copying any skill into another repo.
MIT. See LICENSE.