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-prompt-specialistsgit 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-prompt-specialists)<a href="https://agentmods.dev/skills/anymouschina/tapcanvas/tapcanvas-prompt-specialists"><img src="https://agentmods.dev/badge/skills/anymouschina/tapcanvas/tapcanvas-prompt-specialists/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-prompt-specialists"><img src="https://agentmods.dev/badge/skills/anymouschina/tapcanvas/tapcanvas-prompt-specialists.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 | $0.00045 | $0.01512 |
| Opus 5 | $0.00023 | $0.00756 |
| Sonnet 5 | $0.00009 | $0.00302 |
| Haiku 4.5 | $0.00005 | $0.00151 |
Grade A, and why
tapcanvas-prompt-specialists 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 12d 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 — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.
TapCanvas Prompt Specialists
何时使用
- 主代理需要把图片/视频提示词生产委托给 specialist
- 任务已经具备足够证据，适合进入专门提示词产出阶段
输入证据
- 只传本轮已证实事实
- 推荐包含：目标、关键视觉事实、必须保留、禁止项、对白、时长/节奏上限
- 不要把未确认剧情、角色设定或猜测状态交给 specialist
执行原则
- 是否调用 specialist，由主代理决定
- 不维持固定的后端调用顺序；但当目标是 chapter-grounded 的最终图片/分镜提示词时，应优先使用
image_prompt_specialist，当目标是最终视频提示词时，应组合video_prompt_specialist与pacing_reviewer - handoff 要小而硬，只传最有价值的已证实事实
- specialist 结果必须能回溯到证据，不得成为新事实来源
- 当目标是“优化当前图片节点提示词”时，主代理可以把当前节点的
prompt/systemPrompt/negativePrompt、结果图、参考图和用户本轮明确要求交给image_prompt_specialist，再把 specialist 结果回写原节点 - specialist 负责产出更可执行的视觉提示词，不负责决定是否新建节点；若用户没有要求分叉，主代理应优先更新当前节点而不是复制一份新节点
- 若当前选中输入只是角色卡、三视图或角色参考，而不是已确认场景关键帧，handoff 中必须如实标注为“角色锚点/参考输入”，不得伪装成已确认场景
输出契约
image_prompt_specialist- 最小结果：
{"imagePrompt": string} - 对普通图片优化任务，可只返回
imagePrompt - 对 chapter-grounded / storyboard / keyframe 图片生产，若要额外返回结构化 JSON 编辑视图，最小结果升级为
{"imagePrompt": string, "structuredPrompt": {...}} structuredPrompt是与imagePrompt等价的结构化提示词视图；imagePrompt/ 节点prompt仍是最终执行字段。两者必须可由同一份内容对齐，不要只给其中一个然后假设下游会脑补另一半structuredPrompt至少必须包含：version: "v2"shotIntentspatialLayoutcameraPlanlightingPlancontinuityConstraintsnegativeConstraints
imagePrompt必须是可直接给图片模型执行的最终长提示词，不得退化成一句摘要、标题或营销式短句- 对 chapter-grounded / storyboard / keyframe 场景，
imagePrompt默认应是高信息密度长提示词：优先覆盖并按自然顺序折叠进正文- 时间/天气/光线
- 场景拓扑与空间层次
- 前景 / 中景 / 背景分别有什么
- 画面里有几类主体、谁在左/右/前/后、谁与谁发生什么关系
- 关键道具、机械、建筑、地面、烟尘、纸屑等物理细节
- 机位、焦段感、景别、构图重心、镜头高度、透视关系
- 表情与动作边界
- 风格落点与明确禁止项
- 若证据复杂，优先写得更具体，而不是更短；不要因为“简洁”主动丢失人物数量、位置关系、动作结果、遮挡关系、景深层次或画面主次
- 当参考图只是角色卡/三视图/角色锚点时，必须明确写成“人物外观严格参考图X”，但不能把角色锚点误写成完整场景依据
- 若存在多张参考图，必须在
imagePrompt正文里显式写明图位职责，例如“人物外观严格参考图1，场景构图与冷灰天光延续图2” structuredPrompt推荐增强字段：subjectRelations、environmentObjects、styleConstraints- 推荐长度：通常 300-1200 汉字；场景复杂时可以更长，只要信息仍然可执行、无空话、无重复
- 最小结果：
video_prompt_specialist- 最小结果：
{"prompt": string} - 若仍需保留拍点拆解，可额外给
storyBeatPlan，但它不是执行字段 - 当目标是“最终可执行的视频提示词”或
composeVideo/video节点配置时，所有会影响生成的内容都必须直接折叠进prompt：- 导演意图
- 经典镜头语法借鉴
- 显式动作与结果
- 物理/空间约束
- 禁止漂移项
- 最小结果：
pacing_reviewer- 最小结果：
{"compressionRisk": string, "splitRecommendation": string} - 推荐增强字段：
explicitnessReview、physicsSanityReview
- 最小结果：
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.
- 12d ago First seen · 89 lines · 45 tokens per session scan A f5676ed93efc
tapcanvas-prompt-specialists is a skill published in the GitHub repository anymouschina/TapCanvas (603 stars, last pushed yesterday), licensed MIT. It adds 45 tokens to every session and 1,512 once invoked, about $0.0002 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
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.
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.
short-drama-image-prompts
A guide for writing reusable image prompts for short dramas, covering characters, costumes, locations, props, and different states. It produces copy-ready Markdown prompts but does not create the images.
remake-reel
Analyze a reference video (a reel, montage, or ad the user likes) into an edit blueprint — shot boundaries, music beats, BPM, energy curve, the drop — and rebuild the same structure with the user's own footage. Use when someone says "make it like this video", "remake this reel with my clips", or asks what makes an…
pr-walkthrough
Create a narrated video walkthrough of a pull request with code slides and audio narration. Use when asked to create a PR walkthrough, PR video, or walkthrough video.
listenhub-voice
An end-to-end audio generator that turns text or an image into a finished audio track. It can include sound effects, several speaking voices, or a voice copied from a reference recording.