QwenPaw is a personal AI assistant that runs on a local machine or in the cloud and connects to multiple chat applications. It provides memory, file workspaces, multiple agents, skills, plugins, and integrations with language-model providers and external tools. The catalogue entries are skills that extend its capabilities.
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 agentscope-ai/QwenPaw --skill visual-asset-designgit clone --depth 1 https://github.com/agentscope-ai/QwenPawWrote 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/agentscope-ai/qwenpaw/visual-asset-design)<a href="https://agentmods.dev/skills/agentscope-ai/qwenpaw/visual-asset-design"><img src="https://agentmods.dev/badge/skills/agentscope-ai/qwenpaw/visual-asset-design/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/agentscope-ai/qwenpaw/visual-asset-design"><img src="https://agentmods.dev/badge/skills/agentscope-ai/qwenpaw/visual-asset-design.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00117 | $0.07616 |
| Opus 5.5 | $0.00047 | $0.03046 |
| Sonnet 5.5 | $0.00023 | $0.01523 |
| Haiku 4.5 | $0.00012 | $0.00762 |
Grade A, and why
visual-asset-design 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 1mo 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 — 257 lines — stays where its author put it; the contents beside it link to each section on GitHub.
视觉资产设计
本 skill 提供把角色/场景/道具的创意设定编译为可执行图片生成 Prompt 的领域
知识。它不替代 Project Schema、模型能力限制或用户要求;冲突时遵循:用户
明确要求 > 当前图片模型要求 > 本 skill 默认值。成本合同(一图一 Variant、
生成前去重、required_variant_ids 计划合同)以主 Agent 系统提示中的结构
合同为准,本 skill 只负责"画什么、怎么写"。
1. 参考图布局规范
核心原则
- 类型驱动:角色身份、场景环境与道具的静态设定都采用有主次的多视角参考板:一个较大的主视图建立整体,较小的独立视图补充角度、空间或细节。场景保持无角色空镜,道具聚焦结构与材质;需要表达按时间发生的动作过程时才使用序列关键帧。信息密度服从身份与空间可读性,不默认等大宫格,不为填满版面重复或重叠。
- 用户优先:用户明确要求单图、指定宫格数、指定视角或布局时,按用户要求执行。
角色身份参考图:电影感艺术身份板
角色首次建立视觉身份时,默认生成一张具有高端动画工作室角色研究与艺术书布局感的身份板,而不是标准网格、蓝图、目录或重复 turnaround:
- 画幅使用 Project
settings.aspect_ratio(用户明确指定时以用户为准),不要默认横屏,纯白或柔和米白背景,大面积留白;无环境叙事、无无关道具、无水印。身份性服装与随身装备必须完整保留。 - 版式不对称、优雅且有意失衡,使用多样化图像比例。一个大型、略偏离中心的英雄全身视角作为视觉锚点,周围以干净间距放置较小的独立研究。
- 辅助研究按身份信息价值选择:中性全身、正/侧/背面、坐姿、倾斜、蹲姿、俯视、仰视、富有表现力的肖像;避免多个近似的正面站姿。
- 所有全身、肖像、轮廓和细节研究严格分离:不得重叠、融合、堆叠;不得裁切脸部、隐藏肢体或截断尾巴/身份装备。
- 包含小型轮廓研究区(2–3 个简化黑色轮廓)、小型表情研究区(细微可观察的情绪变化)、小型细节研究区(脸部、发型/毛发、服装结构与关键装备)。
- 文字只保留简约角色 ID 块:名称、角色、核心情绪、视觉标志。手写标签与编辑箭头只在确有帮助时少量使用,不生成长段说明。
- 全部视角必须是同一个角色:相同脸部与比例、发型/毛发、身体比例、服装、装备、固有配色、姿态语言和视觉个性。拟人动物明确双足/四足形态,禁止物种、年龄、体型或服装漂移。
用户明确要求标准设定表、单图或指定宫格时服从用户;仍保持视图独立、信息不重复和身份锁定。
序列关键帧参考图
一张多宫格图,按时间顺序展示主体在一段内容中的关键瞬间:
- 从左到右、从上到下,每格对应一个关键节点或动作瞬间。
- 若该图用于分镜或视频生成参考,每个格子的内部画框必须与目标视频
settings.aspect_ratio完全一致,且必须使用正方形网格(N 列×N 行):把画布按 列×行 切分会让单格比例乘上 行/列,只有列数等于行数时单格才等于目标画幅。格数不是完全平方数时补到下一个完全平方数(2–4 格→2×2,5–9 格→3×3),多余格作为无边框外侧留白,不得拉伸、裁边或用重复内容填空。 - 宫格数量优先取完全平方数:4 格用于简短状态,9 格用于动作密集或完整场景;关键瞬间数量介于两者之间时用 3×3 并接受留白,不要为了填满而增删瞬间。超过 9 格时优先简化背景与单格细节,保持动作轮廓可读。
- 每格应包含:主体动作/状态、环境背景、镜头景别。
- Prompt 中逐格描述。
- 用户要求单图时:单图展示最关键瞬间。
适用场景(不限于剧情):
- 叙事类:情节推进(角色奔跑 → 回头 → 惊讶)。
- 展示类:使用流程(开箱 → 手持 → 使用 → 效果)。
- 动作类:动作序列(准备 → 起跳 → 落地 → 庆祝)。
- 宠物类:行为链(嗅闻 → 追扑 → 进食 → 休息)。
- 旅行类:地点序列(到达 → 探索 → 发现 → 离开)。
- 访谈类:关键瞬间(陈述 → 手势强调 → 反应表情)。
场景/环境参考图:多视角环境研究板
默认在一张图中建立可复用的空间身份,画幅遵循 Project settings.aspect_ratio:
- 使用一个较大的环境全景或广角主视图展示整体空间、主要区域、光影与标志性地标;周围安排比例不同、清晰分离的辅助视图,留白和间距形成阅读层次,不默认使用等大宫格。
- 根据空间需要选择反向视角、侧向视角、俯视空间关系,以及入口、路径、地标、植被、建筑连接处或表面材质的细节。这些是候选项,不要求全部画出;优先用主视图加 1–2 个可明确核对方位的辅助视图建立空间,再按需要补充材质局部。每个视图提供不同信息,兼顾总体与局部,不重复多个近似全景。若辅助视图反复产生空间矛盾,删去冗余机位、收紧视角清单,不在同一张板上继续堆叠方向描述。
- 所有视图必须对应同一个空间和同一时刻的状态:建筑结构、地标位置、路径连接、尺度、固有材质和光照方向保持一致。角度变化不能变成几个独立地点;白天和夕阳作为不同 Variant,并按主图衍生规则制作。
- 先用入口、舞台或其他稳定地标定义空间方位,再逐视图写清机位和朝向;不要把“左侧柜台、右侧窗户”无差别套到所有机位。例如,从入口面向舞台时窗户在画面右侧,从舞台回望入口时同一排窗户应在画面左侧,柜台反之。主视图、反打与俯视图的门窗、通道、家具和采光来源须可相互核对;换机位改变的是画面方向,不是建筑布局。生成后发现辅助视图与主视图矛盾时先修正再供下游使用,不仅凭画面风格相似认定空间一致。
- 场景锚点是环境空镜:在 prompt 中显式声明“空镜、无人物”,不添加具体角色或路人;需要尺度参照时仅使用无身份剪影。角色一致性由角色锚点图负责。
- 用户明确要求单张全景、指定视角、细节重点或宫格时按要求执行,仍保留空间一致性。
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.
- 1mo ago Changed · +46 lines · +4 tokens per session 21842bb046f1
- 1mo ago First seen · 211 lines · 113 tokens per session scan A 707fd212a3c1
visual-asset-design is a skill published in the GitHub repository agentscope-ai/QwenPaw (36,385 stars, last pushed today), licensed Apache-2.0. It adds 117 tokens to every session and 7,616 once invoked, about $0.0005 per session on Opus 5.5. 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 2026-09-09.
Other skills, from other repositories
web-design-engineer
Build or redesign polished browser-rendered visual artifacts with HTML/CSS/JavaScript/React: pages, dashboards, prototypes, slide decks, animations, UI mockups, and data visualizations. Use for visual front-end creation, design-system exploration, design critique, or explicit browser acceptance / QA of a web artifact.…
yida-image-assets
An asset-preparation guide that turns a design's image requirements into an asset manifest for a project. It identifies which page slots need images and records whether assets came from the user, a stock search, or generation.
yida-dashboard
A delivery process for Yida dashboards, meaning screens that show business metrics and charts in one place. It covers data, filters, responsive layouts, loading states, insights, screenshots, and sharing.
yida-cool-card
A tool for Alibaba DingTalk Yida native interactive and pinned cards. These are cards made with DingTalk's built-in components for interaction or pinning, not arbitrary HTML, CSS, or React pages.
yida-density
A set of rules for controlling how much information a custom page shows at once, with compact, comfortable, and spacious layouts. It also defines how the layout changes on phones and how spacing and text sizes stay consistent.
yida-nav-group
A tool for organizing the left-hand navigation of 宜搭 applications, Alibaba’s platform for building business apps. It manages navigation groups, pages, forms, and external links.