interface-design-refinement
/interface-design-refinement
Critique and refine an existing web UI within its authorized visual and functional scope. Use for awkward hierarchy, generic AI-looking pages, inconsistent visual treatment, or explicit UI polish.
- Author
- TMCP-HUB ↗
- Repository
- project-resolver ↗
- Licence
- Apache-2.0
- Last repository change
- 13.11.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 4
No signals detected. Repository changed on 13.11.2026, without references to retired models or APIs.
Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.
What it does and when to use it
Critique and refine an existing web UI within its authorized visual and functional scope. Use for awkward hierarchy, generic AI-looking pages, inconsistent visual treatment, or explicit UI polish.
Original author description
Critique and refine an existing web UI within its authorized visual and functional scope. Use for awkward hierarchy, generic AI-looking pages, inconsistent visual treatment, or explicit UI polish.
How it works
- Purpose
- Goal
- When to use
- Inputs
- Execution contract
- Required outputs
- Procedure
- External source and provenance
- Anti-patterns
- Verification checklist
SKILL.md preview
--- name: interface-design-refinement description: Critique and refine an existing web UI within its authorized visual and functional scope. Use for awkward hierarchy, generic AI-looking pages, inconsistent visual treatment, or explicit UI polish. --- # Interface Design Refinement ## Purpose Improve an existing interface using targeted critique and precise edits, without treating a visual refinement as permission to redesign the product. ## Goal Make the selected page or component clearer, more distinctive where appropriate, and easier to use while preserving the established brand, content, interactions and existing design system. ## When to use Use when asked to critique, polish, simplify, strengthen, adapt, or correct an existing UI or its visual hierarchy. Use `frontend-art-direction` instead for a new marketing concept or an in-scope full visual redesign. Use `design-system-authoring` for tokens and the design contract; `webapp-dogfood-qa` owns required browser-flow evidence. ## Inputs - Exact page/component and authorized change boundary - Existing visuals, `DESIGN.md`/tokens and brand/copy constraints, if available - Surface job and primary user action, where known - Relevant rendered evidence or accessible source; accepted visual claims - Existing implementation stack and available execution capabilities, when implementation is assigned ## Execution contract | Mode | Mandatory capabilities | Claims this mode can support | |---|---|---| | Design critique of supplied artifacts | Model reasoning and supplied source/screenshots | Findings about those artifacts only; not live behavior | | Remote source review | Authorized repository read at the relevant ref | Source-level design defects, not actual rendered geometry | | UI implementation | Access to the actual source and authorized write capability | Source edits in that source; not proof of publication | | Publication, only when assigned | Authorized publication and readback of the exact destination | Published candidate at that destination | | Rendered evaluation, when accepted scope requires it | Capable browser or trustworthy current visual capture for the selected page/state/viewport | Observed appearance; interaction requires actual interaction evidence | Optional: an already installed Impeccable tool, browser inspection, screenshots, and external design references. None is a universal requirement. Reuse existing CI and browser evidence when relevant; do not automatically install Impeccable, run its hooks, open live mode, bypass agent approvals, or add tests. At the cited Impeccable version, live Apply may fall back to a local CLI with approvals and sandboxing disabled when no chat agent is polling; `IMPECCABLE_LIVE_COPY_AGENT=chat` or `off` changes that behavior. Follow the actual execution environment's permission rules and the Owner's already authorized scope. For missing mandatory capabilities before assignment use `ASSIGNMENT_NOT_ADMISSIBLE`; for capability loss during a previously valid execution use `BLOCKED_RUNTIME_DRIFT`. If a required rendered claim is not observed, report `NOT_PROVEN` for that claim; never substitute source-only review. ## Required outputs The in-scope critique or UI delta, the actual design source used, selected improvements, affected files/route when applicable, reused or newly observed evidence, and any unproven claim. If publication is assigned, include the exact published candidate; do not call a local result published. No new design document unless the approved change actually needs one. ## Procedure 1. Bind the requested Owner outcome and any existing visual contract. Establish what may change and what must stay. 2. Inspect only the target and adjacent components necessary to understand its hierarchy, user task, navigation, density, typography, spacing and accessibility. 3. Rank concrete improvements by user impact; distinguish usability problems from a subjective preference for a different aesthetic. 4. Select the smallest appropriate refinement move: clarify content/CTA, distill visual noise, strengthen hierarchy, tune typography/layout, adapt the affected viewport, or harden a relevant edge state. 5. Where available, reuse established tokens, components and accessible behavior. Preserve truthful business content, SEO/semantic structure, keyboard/focus patterns and data interactions. 6. When implementation is assigned, make only the authorized changes. Motion belongs to `purposeful-ui-animation` when motion itself is in scope; no automatic animation or dependency. 7. Check only the claims requiring proof using an admitted mode and available or newly gathered evidence. A rendered conclusion needs visual evidence; a working form needs interaction evidence. Record exact limitations. ## External source and provenance - **Inspiration and workflow reference:** [Impeccable by Paul Bakaus](https://github.com/pbakaus/impeccable), [official docs](https://impeccable.style/docs), [consulted source SKILL.md at commit 687bc381](https://github.com/pbakaus/impeccable/blob/687bc381bc9091bf252c0a05303640561014a425/.agents/skills/impeccable/SKILL.md). License: [Apache-2.0](https://github.com/pbakaus/impeccable/blob/main/LICENSE). - Its critique, polish, distill, typeset, layout, adapt and harden techniques are reference methods, not mandatory commands or a delegated source of Owner authority. Consult the smallest relevant upstream playbook on demand when accessible. No upstream installer, executable, hook, or browser overlay is bundled here. ## Anti-patterns - Replacing an approved identity or factual copy merely to escape a familiar AI aesthetic - Turning a narrow fix into a site-wide redesign, global detector, new tests, or new QA gate - Assuming an upstream preference (font, layout, imagery, motion) overrides the project's Canon - Treating a CSS diff, CI success, or static screenshot as proof of all interactions or responsive states - Automatically executing third-party scripts, live Apply, or bypass-permission paths ## Verification checklist - [ ] The target, applicable design authority and allowed scope are identified. - [ ] Each change serves the requested outcome instead of importing generic visual rules. - [ ] Existing tokens, behavior and content outside scope are preserved. - [ ] Delivered artifacts and any required claim-specific observations have matching evidence. - [ ] Required unavailable evidence is marked `NOT_PROVEN`, not silently claimed. ## Minimal verdict format - Scope and design source - Changes and exact candidate (if any) - Evidence by accepted claim - Remaining limitations - PASS / PARTIAL / FAIL
Versions
The database has no other skills named interface-design-refinement.
Database snapshot: 13.11.2026 · package version 0f565c642bab (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100interface-refinement
Refine an existing interface with named commands (critique, polish, distill, typeset, colorize ...), a product and design context file, and a static detector of generic UI patterns. Use when UI looks generic, cluttered or unfinished. - 95interface-polish
Audit and improve an existing web app's UI and UX while preserving its design language. Use for polish passes, inconsistent typography or icons, cramped panes, weak contrast, awkward controls, snapping menus, and requests to make a shipped interface feel more considered. Covers… - 86workflow-interface-design
Design new web UI or make a substantial visual refinement within the web-development workflow. Use when the visual direction needs design judgment and no explicit or established primary design specialist already owns it. Preserve the actual brief and approved DESIGN.md. - 90ui-slop-score
Review a rendered web or mobile interface and score how generic it looks. Use for UI critique, design review, visual polish, screenshot review, pre-merge checks, and requests to fix AI-looking UI or explain why an interface feels generic. - 100design-audit
Premium UI/UX design audit and refinement skill. Conducts systematic visual audits of existing apps and produces phased, implementation-ready design plans. Use this skill whenever the user asks to audit a UI, improve an app's visual design, make an interface feel more polished… - 100visual-polish-loop
Improve an existing web interface through a bounded screenshot-driven loop: capture a reproducible baseline, fix the highest-impact visual issues, and verify the same states on desktop and mobile. Use when a working UI needs polish or responsive cleanup; do not use for net-new…