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 skills add anymouschina/TapCanvas --skill tapcanvas-visual-focusgit clone --depth 1 https://github.com/anymouschina/TapCanvasWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/anymouschina/tapcanvas/tapcanvas-visual-focus)<a href="https://agentmods.dev/skills/anymouschina/tapcanvas/tapcanvas-visual-focus"><img src="https://agentmods.dev/badge/skills/anymouschina/tapcanvas/tapcanvas-visual-focus/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/anymouschina/tapcanvas/tapcanvas-visual-focus"><img src="https://agentmods.dev/badge/skills/anymouschina/tapcanvas/tapcanvas-visual-focus.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What 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 5.1 | $0.00066 | $0.01142 |
| Opus 5 | $0.00033 | $0.00571 |
| Sonnet 5 | $0.00013 | $0.00228 |
| Haiku 4.5 | $0.00007 | $0.00114 |
Grade A, and why
tapcanvas-visual-focus 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 10d 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.
What it actually says
TapCanvas 视觉重点提取（Visual Focus）
适用场景：
- 用户给了图片（URL 或 DataURL），并希望你提取“视觉重点/关键要素/注意力在哪里/风格要点/拍摄布光要点”。
- 用户的目标来自对话：例如“复现这张图”“做同风格海报”“做分镜/视频风格对齐”“提炼这张图的视觉主轴”。
目标
- 不“看图编故事”：只基于图中可见内容；不确定就标注“推测/不清晰”。
- 把“视觉重点”拆成可执行清单：必须保留 vs 可以替换。
- 给出后续生成/改图可直接用的“提示词骨架”（必要时给英文 prompt + negative prompt）。
工作流（按对话驱动）
- 拿到图片输入
- 优先使用用户给的
imageUrl（http(s) 或相对路径/...）。 - 其次
imageData（data:image/*;base64,...）。 - 如果用户没给图片或给的是
blob:/本地路径，先让用户上传为可访问 URL，或让其提供 DataURL。
- 从对话中抽取“用户想要的视觉重点”
- 用途：复现/改图/海报/视频/风格对齐
- 强调：主体、材质、文字、光线、氛围、配色、构图、镜头
- 忽略：不想要的元素（背景、道具、文字等）
- 输出偏好：要中文总结、要 JSON、要英文 prompt、要负面词等
- 调用图片理解工具（必做）
- 使用
tapcanvas_vision。 - 默认：
modelAlias="gemini-3.1-flash-image-preview"、temperature=0.2。 - 若上游已经提供外部
prompt，直接透传；若未提供，再使用下方推荐模板。 - 若需要固定厂商（避免 auto 路由到其它厂商），传
vendorCandidates（例如仅允许某个厂商）。
建议给 tapcanvas_vision.prompt 的模板（把对话目标填进去）：
“你是资深视觉导演与提示词工程师。请基于输入图片，结合我的目标：<用户目标>，提取我需要的视觉重点。输出严格有效 JSON（不要代码块/多余文字），字段如下：
- summaryZh: 一句话画面摘要
- focusPoints: 数组，每项包含 pointZh（重点）、importance（1-5）、evidence（图中依据/位置特征）、mustKeep（true/false）、canVary（可替换/可变化的方向）
- composition: 构图与镜头（景别/视角/主体位置/景深与焦段倾向；推测需标注）
- lighting: 光线（主光方向、软硬、色温、阴影特征）
- colorPalette: 主色/辅色与氛围
- styleTags: 风格标签数组（摄影/插画/3D/动漫等）
- textInImage: 画面可见文字（不清晰就写不清晰）
- englishPrompt: 可复现英文提示词（只输出提示词正文）
- negativePrompt: 建议的反向提示词（可选）
注意：不要编造不存在的品牌/文字/认证信息；对不确定内容用‘推测/可能’。””
- 把结果变成“可执行”输出
- 如果返回是 JSON：优先按 focusPoints 输出，并补齐“必须保留/可以替换/提示词骨架”。
- 如果返回不是 JSON：按同样结构用中文整理一份（不要臆测）。
推荐输出模板（中文）
- 视觉重点（Top 5-10）：（每条含重要度、必须保留/可替换）
- 必须保留：用于风格对齐/复现的关键要素
- 可以替换：在不破坏风格前提下可变化的要素
- 提示词骨架（用于生成/改图）：中文要点 +（如需要）英文 prompt
- 反向提示词建议（可选）
调用示例（工具参数示意）
tapcanvas_vision参数要点：vendor: "auto"vendorCandidates: ["<可选：限制候选厂商>"]modelAlias: "gemini-3.1-flash-image-preview"temperature: 0.2imageUrl或imageDataprompt: "<按上面模板拼装的任务描述>"
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.
- 10d ago First seen · 73 lines · 66 tokens per session scan A 69fbd3e3571b
tapcanvas-visual-focus is a skill published in the GitHub repository anymouschina/TapCanvas (602 stars, last pushed 2d ago), licensed MIT. It adds 66 tokens to every session and 1,142 once invoked, about $0.0003 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 2026-08-30.
Other skills, from other repositories
slides
Build a Grida slides deck — a .canvas bundle in slides mode whose pages are SVG documents (16:9, one SVG per slide). Use when creating a presentation, pitch deck, slideshow, or talk.
refine-canvas-strokes
Design, implement, review, or troubleshoot safe refinement of editable 2D canvas strokes. Use for handwriting smoothing, automatic text layout cleanup, optional font-guided handwriting normalization, freehand cleanup, geometric shape snapping, diagram cleanup, stroke-style normalization, semantic stroke replacement…
ai-hive-alt-adobe-firefly-workflow
A migration assessment workflow for testing whether AI-HIVE can provide selected creative assets alongside Adobe Firefly and Adobe's editing tools. Adobe Firefly is Adobe's service for generating images and other creative material from instructions.
ai-hive-alt-canva-magic-studio
A migration-testing guide for using AI-HIVE alongside Canva Magic Studio, Canva's tool for template-based design and AI media creation. It tests generated images and videos while leaving Canva's editor and templates in place.
dramake
A production workflow for turning an idea, story, script, storyboard, or existing footage into a short AI drama or film. It covers planning, generation, voice, editing, and quality checks for formats such as vertical and horizontal video.
deck-open-slide-canvas
Locked 1920x1080 canvas deck with React component-level free composition, not bound to a fixed template.