cskwork/supergoal-skill / skills/supergoal
supergoal
UnverifiedRoute objective-driven development, verification, planning, "teach/explain", codebase learning, and skill evaluation through the Supergoal workflow. Use when Supergoal is requested or an autonomous delivery workflow is needed.
reference/engagement.md
View on GitHub# engagement - conversion craft overlay (Expressive surfaces with a primary action)
Load inside the Designer subagent, on top of `reference/taste-skill-v2.md`, when the Design Read names a
primary user action - sign up / buy / book / subscribe / install (SaaS, consumer, commerce, marketing).
Skip for editorial, portfolio, docs, or art-led briefs: forcing conversion patterns there is its own slop.
taste-skill-v2's bans still hold; this only adds the conversion-specific deltas below. Each rule is a
mechanism you build, paired with the slop it replaces.
## Conversion hierarchy
- One dominant primary action per view; give it the most visual weight (size, contrast, isolation). Every
secondary action is quieter. Two equal-weight CTAs is a non-decision the user feels.
- Value-first order: outcome + proof before the form. Repeat the same CTA down a long page; never run two
different primary intents above the fold.
## Hero that converts
- Lead the H1 with one concrete claim - a specific number or named outcome - not a stock photo + generic
tagline. Refines taste's hero rule: still fits the viewport, but the first line earns the scroll. An
unquantified "trusted by thousands" does not convert.
## CTA copy (refines taste's 1-3 word CTA)
- First-person, outcome-named: "Get my audit", "Start my free trial" - not "Submit" / "Sign up" /
"Learn more". Prefer the outcome label even at 3-4 words.
- One friction-reducer line under the button: "Takes 2 minutes. No credit card."
## States as engagement (taste already requires the states; this is the conversion bar)
- Loading: content-shaped skeleton sized to the final content (no layout shift), not a bare spinner.
- Optimistic UI: apply the user-triggered change immediately, reconcile on resolve, roll back on error.
- Empty = onboarding (one purpose line + the single first action); Error = recovery (plain-language cause
+ inline retry, never a dead end or a raw stack trace).
## Trust + social proof (real data only - extends taste's fake-precision ban)
- Put proof where friction peaks: a rating/review/security badge or guarantee adjacent to the CTA, not in
the footer. Prefer named, quantified proof.
- Hard ban: no fabricated counts, no urgency timer that resets every visit. Faked scarcity destroys trust
on the return visit.
## Micro-interaction feedback
- Every interactive element gets immediate hover/active/success feedback tied to a real action. Confirm
each onboarding/checkout step so the user never wonders "did that work?".
- Inline-validate forms at the field on blur with a recovery hint, not one error wall on submit.
- Calm and functional, not gamified (bounce/elastic easing stays banned).
## Motion that guides attention
- Sequenced/staggered entrance to set reading order toward the primary action - a taste hierarchy reason,
never decoration.
- Reveals run off the main thread (native scroll-driven CSS / View Transitions API), with a
`prefers-reduced-motion` fallback. No manual scroll listeners (banned).
## Progressive disclosure
- Defer secondary fields and advanced options behind a reveal; ask the minimum to reach first success.
Map the empty-state-to-first-success path explicitly.
## Accessibility as an engagement lever (WCAG 2.2)
- Visible focus, keyboard-operable everything, no pointer-gesture-only actions, minimum target 24x24 CSS
px (44/48 on mobile). Lowers bounce and support load, not just compliance.
## Critic's engagement pass (judgment, not a gate)
Checked as HIGH/MEDIUM findings, never a deterministic gate rule (engagement is contextual): primary
action obvious; loading/empty/error useful not blank; proof present where the kind expects it; key
interactions have visible feedback; primary CTA copy is outcome-led.
<!-- Adapted from superdesign-skill reference/engagement.md (sibling skill, 2026-06-20). -->
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |