workflow-designer-sample-pages
Skill0 starsChanged 5 months agoMITView on GitHub
Generate self-contained plain-HTML sample pages for a feature from the locked requirement and the design-lead interview results, using exactly one external design toolbox (`ui-ux-pro-max` or `taste-skill`). Duel scope (`samples`) writes candidate pages only; solo scope (`full`) also writes a design-system + token proposal. Output lands under `docs/design-system/sample-candidates/<designer-name>/` in the worktree; never commits. Activate on a dispatch opening with 'Generate sample pages for <feature-name>' or 'Generate design system and sample pages for <feature-name>'.
In this repository74 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- create-bug-issueSkill
- create-enhancement-issueSkill
- create-feature-issuesSkill
- dream-summary-memorySkill
- memory-conventionSkill
- operation-gitSkill
- pattern-architect-api-endpointSkill
- pattern-architect-data-modelSkill
- pattern-architect-deep-moduleSkill
- pattern-e2e-coding-standardSkill
- pattern-engineer-apiSkill
- pattern-engineer-backend-standardSkill
- pattern-engineer-coding-standardSkill
- pattern-engineer-containerSkill
- pattern-engineer-databaseSkill
- pattern-engineer-fastapiSkill
- pattern-engineer-frontend-standardSkill
- pattern-engineer-goSkill
- pattern-engineer-javaSkill
- pattern-engineer-kotlinSkill
- pattern-engineer-nodeSkill
- pattern-engineer-non-functionalSkill
- pattern-engineer-observabilitySkill
- pattern-engineer-pythonSkill
- pattern-engineer-rustSkill
- pattern-engineer-securitySkill
- pattern-engineer-ssrSkill
- pattern-engineer-swiftSkill
- pattern-engineer-typescriptSkill
- pattern-engineer-viteSkill
- pattern-exploratory-testingSkill
- pattern-reviewer-apiSkill
- pattern-reviewer-backend-standardSkill
- pattern-reviewer-coding-standardSkill
- pattern-reviewer-containerSkill
- pattern-reviewer-contractSkill
- pattern-reviewer-databaseSkill
- pattern-reviewer-fastapiSkill
- pattern-reviewer-frontend-standardSkill
- pattern-reviewer-goSkill
- pattern-reviewer-javaSkill
- pattern-reviewer-kotlinSkill
- pattern-reviewer-nodeSkill
- pattern-reviewer-non-functionalSkill
- pattern-reviewer-observabilitySkill
- pattern-reviewer-pythonSkill
- pattern-reviewer-rustSkill
- pattern-reviewer-securitySkill
- pattern-reviewer-ssrSkill
- pattern-reviewer-swiftSkill
- pattern-reviewer-test-coverageSkill
- pattern-reviewer-typescriptSkill
- pattern-reviewer-viteSkill
- pattern-test-coverageSkill
- principle-engineer-tddSkill
- scaffold-projectSkill
- workflow-architect-interviewSkill
- workflow-design-interviewSkill
- workflow-e2e-authorSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
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
- Distinctive: line 19
Do NOT activate when:
- Very specific: line 41
Read docs/product-requirement-document/<feature-name>/requirement.md in the worktree, plus the sibling critical-path and glossary files.
- Very specific: line 53
Write one self-contained .html file per sampled surface under:
- Very specific: line 65
If more than one page, add an index.html linking the set with a one-line note on what each page demonstrates.
- Distinctive: line 79
One coherent direction per run: Commit to a single direction rather than hedging across variants — the duel exists so the human compares …
- Distinctive: line 80
Real content is the demo: Use the requirement's actual entities, labels, empty-state copy, and numbers.
≈ “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
workflow-designer-sample-pages
Turn a locked requirement plus the design-lead's interview results into browsable, self-contained plain-HTML sample pages a human can open from disk and judge. Run by a designer teammate dispatched from /deep-dive-feature after the design interview lands — either as one of two duelling designers (the human votes the winner) or as a solo designer (no vote).
This skill never commits and never pushes — it writes candidate files only, inside the designer's own directory under docs/design-system/sample-candidates/. Moving the winner into docs/design-system/samples/ and committing is the design-writer's job downstream.
When to activate
Activate this skill whenever:
- The dispatch prompt opens with
Generate sample pages for <feature-name>— duel scopesamples: candidate pages only. - The dispatch prompt opens with
Generate design system and sample pages for <feature-name>— solo scopefull: candidate pages plus a design-system + token proposal (proposal.md).
Do NOT activate when:
- The unit of work is the design interview itself (
workflow-design-interview—design-lead's lane). - The unit of work is publishing
docs/design-system/*artifacts or moving the winning samples (workflow-writer-publish-design— the writer's lane). - The dispatch asks for production frontend code (the
engineer's lane).
Inputs from the dispatch
<feature-name>— resolves the requirement path.{worktree_path}— every read and write targets this directory.- The toolbox — exactly one of
ui-ux-pro-max:ui-ux-pro-max/taste-skill:taste-skill(fully-qualified, they live in separate plugins). - Your teammate name (e.g.
designer-pro-max,designer-taste) — names your output directory. - The scope —
samples(duel) orfull(solo), implied by the trigger phrase.
Workflow
1. Load the named toolbox
Load the single toolbox skill the dispatch names. If it is not installed/available, HALT and surface the missing plugin to whoever dispatched you — do not improvise a direction from your own taste (the no-toolbox fallback is design-lead's, not yours).
2. Read the requirement
Read docs/product-requirement-document/<feature-name>/requirement.md in the worktree, plus the sibling critical-path and glossary files. Note the entities, flows, and user-visible labels — your pages use the product's real content, never lorem ipsum.
3. Pull the interview results from design-lead
Send SendMessage(to=design-lead) requesting the locked interview results: the visual-language decisions (brand/tone, color philosophy, typography, spatial rhythm, motion posture, platform priority, accessibility targets) and the surface + navigation inventory. Wait for the reply before designing — these are constraints your toolbox expresses within, not suggestions. Also read any existing docs/design-system/ in the worktree; an extension honors the locked system.
4. Pick 1–3 representative surfaces
From the surface inventory and critical paths, pick the 1–3 surfaces most central to the requirement's critical path (typically the primary working surface, plus the app shell / nav context it sits in). A duel is judged on direction, not coverage — do not build every page.
5. Generate the sample pages
Write one self-contained .html file per sampled surface under:
{worktree_path}/docs/design-system/sample-candidates/<your-teammate-name>/Hard constraints per file:
- Plain HTML with one inline
<style>block — no build step, no framework, no external CSS file. A Google Fonts<link>is acceptable; the page must still be legible without it. - Renders correctly opened directly via
file://. - Semantic elements and the accessibility floor from the interview (contrast, visible focus, tap-target minimum) — non-negotiable.
- At most a few lines of vanilla JS, only to demo an interaction state.
- If more than one page, add an
index.htmllinking the set with a one-line note on what each page demonstrates.
Stay strictly inside your own directory — never touch another designer's candidates, the design-system docs, or production code.
6. Solo scope only: write the proposal
At scope full, also write proposal.md in the same directory: the design-system + token proposal your pages embody — style family, palette with concrete values, font pairing, type scale, spacing scale, radius / shadow / motion tokens, and a short rationale per group. This is input for design-lead to review with the user and fold into the design-writer payload; it is a proposal, not a published artifact.
7. Report and stop
Report back to whoever dispatched you: the absolute file paths written (so the human can open them in a browser) and a one-paragraph statement of the design direction (what the toolbox led you to and why it fits this product). Do not commit, push, tick any checklist, or message the user directly — the orchestrator surfaces the candidates and runs the vote.
Best practices
- One coherent direction per run. Commit to a single direction rather than hedging across variants — the duel exists so the human compares two committed directions.
- Real content is the demo. Use the requirement's actual entities, labels, empty-state copy, and numbers; a sample page with fake content tests nothing.
- Show state, not just layout. Where the critical path implies it, include a filled state, an empty state, or a validation state — cheap in plain HTML, decisive in a vote.
- Tokens visible in the source. Express colors / spacing / type as CSS custom properties at the top of the
<style>block, named like design tokens — it makes the winning direction trivially liftable intotokens.md.
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.

