Skip to content

tw93/Waza / plugins/waza/skills/ui

Produces distinctive production UI and screenshot-grounded visual polish. Use when building or restyling pages, components, or typography. Not for backend logic or data pipelines.

references/mode-screenshot-iteration.md

View on GitHub
# Screenshot Iteration Mode

Loaded from `ui` when the user supplies a screenshot of a rendered surface and wants it improved against that evidence.

Activate when the user sends a screenshot or image alongside a complaint ("这里很丑", "这个不对", "fix this", "looks wrong"). The existing product is the direction. Skip the five-question direction lock.

**Flow:**

1. Read the screenshot. State the problem in one sentence: what specifically looks wrong (spacing, contrast, alignment, typeface, color, density, hierarchy). Preserve the user's negative label when it is diagnostic; do not translate "丑", "乱", "不清晰", or "怪" into vague "make it modern" language.
2. Lock the repair contract before touching code: `target` names the surface and viewport/state allowed to change; `preserve` names the adjacent surface, content, and product behavior that remain unchanged; `evidence` names the current render and any available reference renders that will prove the result. Ask only when two plausible targets would produce materially different behavior; otherwise act on the strongest screenshot evidence.
3. If the user provides a reference screenshot, older version, or "this one is good" example, compare current vs. reference and name the visual deltas before choosing a fix.
4. If the diagnosis is a genuinely unfamiliar UX problem and current product evidence remains underdetermined after reading references, source, and sibling components, spend one round surveying how 2-3 mature products in the same category solve it before writing code. Cite one concrete decision from each. Skip the survey when existing evidence settles the interaction or the fix is cosmetic.
5. Find the responsible code: grep for the component name or class, read the actual file. Do not rely on memory or assumptions about file location.
6. Apply the minimal fix. For existing products, try material/opacity, geometry, spacing, typography, or text-fit adjustments before redesigning the surface.
7. If the complaint says "still" or exposes alignment, sizing, conditional rendering, animation, or shared-component inconsistency, freeze a minimal visual matrix before verification: affected sibling surfaces; web breakpoint sides or native minimum width, minimum height, their combination, and normal size; default, selected/focused, loading, empty/error, and compact/localized states that the change can reach; and before/during/after plus cold/warm paths for transitions. Derive one shared invariant or token where the same defect repeats instead of tuning screenshots independently. When the user is tuning one number by eye (width, size, spacing) and the second value did not land, stop iterating single values: render three labelled candidates side by side at the complained viewport and let the user pick. Container width is one token, taken from a sibling surface when one exists, before any block-level cap exists.
8. Verify the matrix in a browser, native app, screenshot tool, or rendered artifact, then re-check every `preserve` boundary. A final-state screenshot does not prove copy feedback, touch behavior, first paint, or transition stability. If rendering is available, finish the loop yourself and report the checked states. If the host cannot render, say that explicitly and hand off the shortest exact view the user should check. Alignment and spacing complaints close with numbers: measure the two boxes' geometry (`getBoundingClientRect` or a screenshot ruler) before and after, and report both sets plus the viewport width; without numbers the reply does not say "aligned". The reply prints `checked N / changed M / left K` over the consumers found in step 7; a reply that changes one view without that line is the failure this step exists to stop.

**Calibration rules:**
- The user's screenshot is the strongest design brief in the turn. Keep it visible in the reasoning until the fix is done.
- The real running product is the oracle. Product pages, app screenshots, release pages, and current UI state override generic style instincts. Render comparison candidates on the product's actual background with adjacent components and state preserved; vary only the property being compared.
- Do not flatten specific taste feedback into generic UI adjectives. "More premium" is not a diagnosis; "caption baseline drifts above the Chinese line" is.
- If the screenshot exposes a deterministic runtime regression or a broken coded render, state, or timing path, route to `/hunt` and preserve the visual evidence. Generated image assets remain in `mode-generated-asset.md` for both taste and output defects, including malformed labels, mattes, and halos.

**Native screenshot handoff.** For native apps, once you have proven the app builds, runs, and can reach the target view, do not spend repeated cycles fighting focus, window ordering, or coordinate-click automation just to capture final visual proof. Make one bounded automation attempt. If it is flaky, name the exact screen and ask the user for the screenshot to iterate against. This is a visual QA boundary, not a substitute for build/run verification.

**Boundary**: if the fix requires changing 3 or more components, or if it reveals a direction problem rather than a specific bug, pause and run the full direction lock before continuing.

**Redesign priority order** (when reworking an existing UI rather than building from scratch): font replacement first, then color cleanup, hover/active states, layout and whitespace, replacing generic components, adding loading/empty/error states, and typographic polish last. This order maximizes visual lift while minimizing the blast radius of each pass. Full rules, common traps, and absolute CSS bans all live in `references/design-reference.md`.
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall