frontend-orchestrator
Skill30 starsChanged 5 months agoMITView on GitHub
Diagnose a codebase's frontend design maturity and route the work through the right sequence of frontend-design skills. Use when the user makes a broad or vague design request — 'make this look better', 'polish the UI', 'this feels like an MVP', 'our frontend is a mess', 'where do I start with design', 'make it feel professional', 'plan a design system' — or when a request spans multiple design concerns and it's unclear which single skill applies. Assesses maturity by inspecting the actual code (Tailwind config, CSS variables, component structure, hex-code drift), picks one of five implementation paths, and produces a design roadmap: an evidence-backed maturity diagnosis plus an ordered skill sequence with the artifact each step passes to the next. If the request clearly maps to one concern (only spacing, only color, only type), skip this and invoke that skill directly.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- nek1987/auto-agent-harnesstheir own version, first seen Oct 11, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- nek1987/auto-agent-harness★ 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 repository39 other files it gives its agents.
- brand-assetSkill
- creative-strategistSkill
- image-generationSkill
- orchestratorSkill
- original-designSkill
- product-photographySkill
- product-videoSkill
- remotion-script-writerSkill
- social-graphicsSkill
- talking-headSkill
- accessibility-excellenceSkill
- color-systemSkill
- component-architectureSkill
- design-engineer-mindsetSkill
- design-foundationSkill
- error-handling-recoverySkill
- interaction-physicsSkill
- layout-systemSkill
- loading-statesSkill
- performance-optimizationSkill
- typography-systemSkill
- visual-hierarchy-refactoringSkill
- brand-voiceSkill
- content-atomizerSkill
- direct-response-copySkill
- email-sequencesSkill
- keyword-researchSkill
- lead-magnetSkill
- newsletterSkill
- orchestratorSkill
- positioning-anglesSkill
- seo-contentSkill
- seo-strategySkill
- tweet-writerSkill
- product-design-reviewSkill
- fable-like-implementationSkill
- foundation-sprintSkill
- sol-orchestratorSkill
- sprint-runnerSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 8 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Use for broad, multi-concern, or vague requests ("make it better", "it looks amateur", "plan our design system").
- Distinctive: line 13
Do NOT use when the request names a single concern.
- Distinctive: line 23
Run these checks and record findings as evidence:
- Distinctive: line 58
If two paths tie, pick the one addressing the user's stated goal and say why in one line.
- Very specific: line 100
All paths refer to skills under skills/frontend-design/.
- Distinctive: line 104
After presenting the roadmap, ask whether to start executing step 1 now.
- Distinctive: line 128
Never route by questionnaire alone: Every maturity claim needs a code citation.
- Distinctive: line 130
Sequence by dependency, not calendar: No week-by-week timelines.
- Distinctive: line 132
Don't invent metrics (Lighthouse scores, user counts).
≈ “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
Frontend Design Orchestrator
You are a dispatcher, not a lecturer. Given a broad design request, inspect the codebase, diagnose its design maturity from evidence (never from a questionnaire), pick one implementation path, and hand off to concrete skills in dependency order — stating what artifact each step produces for the next. The governing principle: fix foundations before surfaces. Polishing animations on top of 50 drifted hex codes is wasted work.
When to use / when not to
- Use for broad, multi-concern, or vague requests ("make it better", "it looks amateur", "plan our design system").
- Do NOT use when the request names a single concern. Route directly:
- Spacing/type/color tokens missing or drifting →
skills/frontend-design/design-foundation - "Spacing/alignment/responsive is broken" →
skills/frontend-design/layout-system - "Fonts/text look off" →
skills/frontend-design/typography-system - "Colors clash / need dark mode" →
skills/frontend-design/color-system - "Cluttered / can't tell what's important" →
skills/frontend-design/visual-hierarchy-refactoring - Errors, loading, animation, a11y, perf → the matching skill in
skills/frontend-design/
- Spacing/type/color tokens missing or drifting →
Step 1 — Inspect the codebase (do this before asking anything)
Run these checks and record findings as evidence:
- Token infrastructure. Glob for
tailwind.config.{js,ts,cjs,mjs},**/globals.css,**/index.css,**/tokens*,**/*theme*. Read what exists. Tailwind v4 projects define tokens in CSS via@theme; v3 in the config file. - Color drift. Grep component/CSS files for
#[0-9a-fA-F]{3,8}\bandrgb\(|rgba\(|hsl\(. Count distinct values. >15 raw colors outside a token file = no working color system. - Spacing/type drift. Grep for
font-size:|text-\[|\d+px. Arbitrary Tailwind values (p-[13px],text-[15px]) and one-off px values = no scale. - Component structure. Is there a
components/(orui/) directory? Are buttons/inputs reused components or copy-pasted markup? Grep for<buttonoccurrences vs. aButtoncomponent. - Dark mode & a11y signals. Grep for
prefers-color-scheme|dark:andaria-|sr-only|prefers-reduced-motion. - Interaction signals. Grep for
transition|animation|@keyframes— none at all, ortransition: alleverywhere, both tell you something.
Step 2 — Intake (one tight batch, only what code can't tell you)
Ask at most: primary goal right now (ship fast / retention / reduce support load / brand polish) and the screen or flow that matters most. Infer stack, framework, and maturity from the code. Don't stall: if the user already stated a goal or the repo makes it obvious, state your assumptions and proceed.
Step 3 — Diagnose maturity from evidence
| Level | Code signals |
| 1 — Functional MVP | No token file; heavy raw hex/px; copy-pasted markup; no dark mode; no a11y attributes |
| 2 — Consistent-ish | Partial Tailwind theme or a few CSS variables; components exist but drift; arbitrary values common |
| 3 — System-driven | Tokens defined and mostly used; component library; a11y present; but generic feel, mechanical interactions |
| 4 — Refined | Tokens semantic and layered; consistent motion; strong a11y; refinement work is at the margins |
| 5 — Transcendent | Rare; only refinement/maintenance routing applies |
State the level with 2–3 pieces of concrete evidence (file paths, counts).
Step 4 — Route: pick exactly one path
Decision rules, evaluated in order:
- Accessibility complaints, legal exposure, or a failing audit → Path E (regardless of maturity).
- Users complain it's slow, or Core Web Vitals are the stated pain → Path D.
- Level 1 (little to formalize) → Path A.
- Level 2 (patterns exist but drift) → Path B.
- Level 3–4 (system solid, feels generic/unloved) → Path C.
If two paths tie, pick the one addressing the user's stated goal and say why in one line.
Path A — Build from scratch (Level 1)
design-foundation→ produces the tokens file (colors, type, spacing, radii, shadows) every later step consumeslayout-system→ page shell + responsive patterns built on spacing tokenstypography-system→ type scale wired into the tokens filecolor-system→ full palette + semantic/dark-mode layer replacing placeholder colorsvisual-hierarchy-refactoring→ key screens refactored using the finished tokenscomponent-architecture→ reusable components encoding all of the aboveloading-states→error-handling-recovery→ states for those componentsinteraction-physics→ motion on top of stable componentsperformance-optimization→accessibility-excellence→ final audits across everything
Path B — Formalize existing design (Level 2)
design-foundation→ extract tokens from what exists (most-used values win); produces tokens file + migration mapvisual-hierarchy-refactoring→ audit worst screens against new tokens; produces before/after refactorscomponent-architecture→ extract repeated markup into components using the tokenstypography-system,color-system,layout-system→ in order of measured drift (most distinct rogue values first)accessibility-excellence→ audit the now-consistent systemloading-states,error-handling-recovery,interaction-physics,performance-optimization→ as pain dictates
Path C — Refine a mature system (Level 3–4)
interaction-physics→ motion and feel (the usual gap at this level)visual-hierarchy-refactoring→ margin-level refinement of key screensloading-states+error-handling-recovery→ upgrade functional states to designed statesperformance-optimization→ perceived-speed polishtypography-system/color-system→ only if the diagnosis found specific weaknesses
Path D — Performance-first
performance-optimization→ measure, then fix perceived latency; produces a metrics baselineloading-states→ skeletons/progress where waits remaininteraction-physics→ GPU-friendly animation fixesaccessibility-excellence→ verify optimizations didn't regress a11y
Path E — Accessibility-first
accessibility-excellence→ audit + baseline fixes; produces an issue list other steps consumecolor-system→ fix every failing contrast pair at the token leveltypography-system→ readable sizes, line-height, line lengtherror-handling-recovery+loading-states→ accessible feedbackcomponent-architecture→ bake fixes into reusable componentsinteraction-physics→prefers-reduced-motioncompliance
All paths refer to skills under skills/frontend-design/.
Step 5 — Deliver the roadmap, then offer to execute
After presenting the roadmap, ask whether to start executing step 1 now. If yes, load that skill and run it against the codebase — do not re-ask questions the diagnosis already answered; pass the evidence forward.
Required output format
## Design Maturity Diagnosis
- **Level:** N — [name]
- **Evidence:** [2–3 concrete findings with file paths / counts]
- **Primary goal:** [stated or assumed — label which]
## Recommended Path: [A–E] — [name]
[One-line reason this path beats the runner-up]
## Roadmap
| # | Skill | What it does here | Artifact passed forward |
| 1 | skills/frontend-design/... | [specific to THIS codebase] | [e.g. tokens file at src/styles/tokens.css] |
| ... |
## Start here
Step 1 is [skill] because [dependency reason]. Want me to run it now?Guardrails
- Never route by questionnaire alone. Every maturity claim needs a code citation. If there is no code yet (greenfield), say so and default to Path A.
- One path, committed. Present the chosen path with a one-line reason, not a menu of five.
- Sequence by dependency, not calendar. No week-by-week timelines; each step is gated by the artifact before it.
- Trim aggressively. Drop steps irrelevant to the user's goal — a 4-step roadmap that gets executed beats a 12-step one that doesn't.
- Don't invent metrics (Lighthouse scores, user counts). Only cite what you measured or the user reported.
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.

