Using Design Skills

using-design-skills

You decide which design skills a task needs, in what order, and who executes each stage, then defer to them. Member roster, role tables, and executor details live in references/design-roles.md — read it at pipeline composition time; never route from memory of it.

BYPASS

• Backend-only / non-UI work · one-line CSS or copy tweak (inline; at most consult DESIGN.md) · the user names ONE member skill for a single-skill-sized task → go direct.
• Loop-shaped non-design work (audits, plan→build) → using-workflows.

TRIGGER

Any deliverable whose value is visual or interface-shaped: pages, screens, HTML reports/diagrams/plans, artifacts, dashboards, motion, mockups, module/API interface design.

QUESTIONS — compose the pipeline (answer in order)

Q1 Base deliverable — who owns the output? Diagram → diagram-design for house-style consistency / mermaid / draw.io / PNG-SVG export, else the HTML trio · report page → html · plan page → html-plan · web/mobile → target stack · Artifact → artifact file · data file → data-report · module/API/domain → Pipeline D · token work → DESIGN.md itself · mockup-only → imagegen output · "would this work?" → prototype, alone, no loop.

Q2 Variant. Existing interface → audit-first (authority critiques the CURRENT render before rebuild). New + ambitious + imagegen available → image-first (imagegen ⇒ image-to-code ⇒ loop); ordinary task → direct build, say so when skipping imagegen.

Q3 Orthogonal specialists (additive). Charts from a data file → data-report; charts inside a page → apply charting best practice inline, since skills-lock.json holds no dedicated chart skill. Artifact → artifact-design, a skill bundled with the runtime rather than the lock; mandatory before publishing, and skipped explicitly if it does not resolve this turn. Motion or gesture → apple-design. Agency-grade type, spacing, and shadow detail → high-end-visual-design. Constraint sets load inline.

Q4 Executor per stage — the executor table in references/design-roles.md. Direction runs INLINE; imagegen/image-to-code → agent-tmux codex persistent worker; non-trivial build → agent-tmux claude persistent worker; every review → FRESH headless one-shot, never the author.

Q5 Verifier. Render in scope → the visual quality loop below (full Gate 0 screenshot evidence). DESIGN.md-only → document-only evidence (file hash/commit, token completeness, contrast math). Pipeline D → non-visual loop: design comparison judged against codebase-design depth criteria by a fresh reviewer, verdict still PASS/BLOCK. prototype → none.

SELECT — discover live, then bind roles

1. jq -r '.skills|keys[]' ~/.agents/.skill-lock.json is the installed roster; cross-check against THIS turn's active available-skills listing, because an installed directory that is not loaded is not invocable. A member gated disable-model-invocation: true (check head -8) is reached by reading its SKILL.md inline, never via Skill().
2. Exactly ONE direction authority for a visual pipeline (ZERO for Pipeline D / prototypes / trivial tweaks). Never stack two — they fight. An unknown skill is never auto-promoted to authority.
3. A prior member that is absent or uncallable → STATE the absence and the substitute in your plan; an empty role → degrade explicitly. Never substitute silently.
4. Read each selected skill's SKILL.md before its stage runs.

DEFER — cross-stage contracts

• DESIGN.md: the direction stage WRITES it (plain Read/Write — the design-md helper skill was removed); every later stage and every worker prompt READS it. Workers have no chat memory — never carry direction in chat only.
• Dispatch: tmux dispatches route through using-tmux-agent-tools; worker prompts from delegation-templates; tiers per ~/.agents/rules/model-dispatch.md §5, verification §7. Fanout/dialogue need the user's exact authorization.
• Loop topology (fixed): the BUILDER is a persistent worker — BLOCK findings return to the SAME builder (worker-reuse protocol). Each REVIEWER is a fresh headless one-shot, never the builder, never reused.
• Worker-prompt addendum: "DESIGN CONTEXT: read {repo}/DESIGN.md first; conform to its tokens. SKILL: read and follow {member-SKILL.md} first."
• Diagram house defaults (these override a member's opt-in gate, never its mechanism): a diagram whose story IS direction — data flow, architecture, dependency, deployment — ships motion by default. Request diagram-design mode loop with the Flow-token primitive (one aria-hidden token, cycle ≥3s, no playback controls); escalate to reveal + Path draw only when the ORDER of hops carries meaning, and to step only for teaching. Every other diagram type (ER, quadrant, bar, Venn) defaults to none — no flow to point at means motion is decoration. The static-first contract is NOT relaxed: with JS off, and in reduced-motion, print, and every export, the complete labeled figure is visible.
• Format: HTML first. Every diagram and visual deliverable is a self-contained .html; svg/png are EXPORTED from it, never hand-authored. Markdown (a mermaid fence) requires the user asking for it or a text-only consuming tool — never the agent's own guess that "the reader probably cannot see images".

The quality loop — no pipeline ends at "built"

Gate 0, fail-closed: the reviewer RECORDS build identity (hash/commit), artifact path/URL, screenshot path(s), viewport(s), full-page/uncropped. Desktop AND mobile for responsive deliverables. Missing/stale/cropped evidence → UNCONFIRMED → overall BLOCK; never PASS on code-only review. (Document-only exception: DESIGN.md-only deliverables record file hash/commit instead — the instant any render exists, full contract resumes.)

Rubric, in order: 1. THE SLOP TEST — the authority's audit rubric; generic gradient-hero / emoji-bullet / card-grid sameness = BLOCK regardless of correctness. 2. Prose discipline — no filler, scannable hierarchy. 3. DESIGN.md conformance. 4. The stage's frozen ACCEPTANCE criteria.

What counts as BLOCK: only a rubric item above or a frozen ACCEPTANCE criterion. Accessibility polish (hit-area, contrast, aria names, focus ring, skip link) that no ACCEPTANCE line names is a NOTE — list it, do not BLOCK on it, do not open a fix round for it. A reviewer that returns only NOTEs returns VERDICT: PASS.

Report contract: per-item PASS / FAIL / UNCONFIRMED + one-line reason, findings tagged persistent/new/regression/evidence-gap, all blockers in ONE pass, final line exactly VERDICT: PASS or VERDICT: BLOCK, quoted verbatim in your report. A report missing the VERDICT: line is a format defect of that reviewer: re-ask the SAME reviewer for the one line — never spawn another reviewer to obtain it.

Rounds: BLOCK → same persistent builder fixes → fresh reviewer re-audits. Hard cap two fix rounds, then route by CAUSE: same-root failures

• wrong-direction signals (judgment-rubrics.md §4) → back to direction · evidence-gap → repair evidence path, re-audit · new/regression → triage separately · otherwise → stop and ask the user with the failure trail. PASS → write the final tokens into DESIGN.md, report with the evidence bundle. "Built and audited: PASS" is a completion claim; "built" alone is not.

Polish/Operate mode — human checkpoint first: the first rendered build goes to the USER (URL/screenshot, desktop + mobile) BEFORE any reviewer is spawned. Direction is the user's call; the reviewer only checks the direction the user accepted. One reviewer round, not a loop.

NOT-FOUND

A needed role has no installed member → degrade the pipeline explicitly and say what was skipped — never fake a stage. Member contracts: each skill's own SKILL.md · roster/roles/executors: references/design-roles.md · worker mechanics: tmux-agent-tools · prompts: delegation-templates.