Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/yangchao228/loop-builder/ui-visual-match-replicatornpx skills add yangchao228/loop-builder --skill ui-visual-match-replicatorgit clone --depth 1 https://github.com/yangchao228/loop-builderWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00093 | $0.01189 |
| Opus 5 | $0.00046 | $0.00594 |
| Sonnet 5 | $0.00019 | $0.00238 |
| Haiku 4.5 | $0.00009 | $0.00119 |
Grade A, and why
ui-visual-match-replicator scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 148 lines — stays where its author put it; the contents beside it link to each section on GitHub.
UI Visual Match Replicator
Role
Act as a senior frontend visual replication engineer. Recreate the target UI with disciplined fidelity; do not act as a free-form designer.
Use this skill to run a controlled visual match loop:
target image -> visual breakdown -> implementation -> rendered screenshot -> Top differences -> fix Top 3 -> repeat -> final gap report
For the full loop contract and prompt skeleton, read references/visual-match-loop.md when starting a real task.
Input Gate
Before editing code, identify the required inputs:
- Target visual: screenshot, image path, Figma export, current accepted demo, or target URL screenshot.
- Working directory or project surface.
- Delivery form: static demo, existing app page, component, H5, mini program, or web page.
- Primary viewport and secondary viewport.
- Allowed modification scope.
- Forbidden actions.
If the target visual is missing or unreadable, stop and ask for it. Do not infer the design from text alone.
If the working directory is missing but the current thread has a clear repo, use that repo and state the assumption.
Default safe assumptions:
- Build a static demo first when the user asks for visual exploration or has not yet accepted implementation into production.
- Modify only UI files, styles, and local visual assets unless the user explicitly allows business logic changes.
- Do not introduce a UI framework or new dependency without confirmation.
- Do not submit, publish, deploy, or change production configuration.
Required First Step
Start with read-only visual and repo analysis. Do not write code yet.
Visual breakdown must cover:
- Page structure.
- Component tree.
- Color tokens.
- Font sizes, weights, and line heights.
- Spacing system.
- Border radii and shadows.
- Backgrounds, translucency, gradients, and blur.
- Icon and image handling.
- Responsive or viewport scaling strategy.
Repo analysis must cover:
- Existing demo, static, docs, public, assets, and page/component directories.
- Existing visual assets that can be reused.
- Existing scripts for dev server, build, type check, and tests.
- Dirty worktree state, without reverting user changes.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 148 lines · 93 tokens per session scan A 947073e8015f
ui-visual-match-replicator is a skill published in the GitHub repository yangchao228/loop-builder (5 stars, last pushed 1mo ago), licensed MIT. It adds 93 tokens to every session and 1,189 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
no-negative-echo
Prevent 此地无银三百两式 residue: finalize artifacts without echoing rejected session-only alternatives into labels, metadata, commits, PRs, or handoffs. Use after corrections or discarded proposals. Not for ordinary deletion, deprecation, migration, or exclusions materially required for safety, accuracy, compatibility…
female-portrait-director
Generate, visually expand, optimize, diagnose, and route structured AI image prompts for adult female portraits. Use for lifestyle, curve-focused, fashion, oriental, fantasy, realism, beauty, CCD, low-key cinematic, studio, sport, travel, and e-commerce portrait requests; onboarding or usage help; parameter…
gc-still-image-motion-director
Analyze still images and design restrained, image-specific motion for image-to-video generation. Use when the user provides one or more images, screenshots, posters, collages, illustrations, editorial graphics, or an image folder and asks how the image should move, which elements must stay fixed, whether it should…
ian-xiaohei-scenes
生成“小黑 + 真实物件 + 物理动作 + 留白叙事”的中文配图。默认输出 16:9 正文配图，用于互联网打工人共鸣图、AI 时代职场焦虑、程序员/产品/创作者处境和正文观点隐喻图；遇到“彩蛋模式 / 长卷故事图 / 超横版 / 个人经历 / 项目复盘 / 产品演化 / 成长路径”时，输出小黑长卷故事图。标准模式默认 #FFFFFF 纯白背景；彩蛋长卷模式使用高级近白背景和一条真实物件人生线。样例是高质量模板母版和出图质量标尺；必须对齐其比例、留白、动作清晰度和叙事关系，但不能复刻其物件组合、空间拓扑、小黑姿态或标签位置。.
photo-revival
Turn ordinary photos or everyday snapshots into poetic hand-drawn illustrations with large white-paper negative space, vivid localized color, subtle collage accents, and tiny handwritten captions. Use when a user wants a real photo reimagined as an illustration rather than filtered, beautified, or photorealistically…
advisor-gate
A procedure for consulting a frontier-class external reasoning model (extended-thinking tier) on a decision that matters — and for accepting its answer afterwards. Generates the consultation prompt, then audits the reply. Vendor-neutral: any top-tier model will do. Not for routine work sent to a cheap model — that is…