Canghe Comic
freestylefly/canghe-skills
Knowledge comic creator supporting multiple art styles and tones.
Guides writing and repairing image prompts for character, scene and prop reference sheets, including layout, camera language and continuity across variants.
SKILL.md written in Chinese; this summary is our English description.
$ npx skills add agentscope-ai/QwenPaw --skill visual-asset-design -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install agentscope-ai/QwenPaw visual-asset-design --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/agentscope-ai/QwenPaw.git skills-src && mkdir -p .claude/skills && cp -r skills-src/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design .claude/skills/visual-asset-design && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "visual-asset-design" agent skill from https://github.com/agentscope-ai/QwenPaw/tree/main/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design into .claude/skills/visual-asset-design/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "visual-asset-design", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/agentscope-ai/QwenPaw/tree/main/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-designType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add agentscope-ai/QwenPaw --skill visual-asset-design -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install agentscope-ai/QwenPaw visual-asset-design --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/agentscope-ai/QwenPaw.git skills-src && mkdir -p .agents/skills && cp -r skills-src/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design .agents/skills/visual-asset-design && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "visual-asset-design" agent skill from https://github.com/agentscope-ai/QwenPaw/tree/main/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design into .agents/skills/visual-asset-design/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "visual-asset-design", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add agentscope-ai/QwenPaw --skill visual-asset-design -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install agentscope-ai/QwenPaw visual-asset-design --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/agentscope-ai/QwenPaw.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design .cursor/skills/visual-asset-design && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "visual-asset-design" agent skill from https://github.com/agentscope-ai/QwenPaw/tree/main/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design into .cursor/skills/visual-asset-design/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "visual-asset-design", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/agentscope-ai/QwenPaw.git --path plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add agentscope-ai/QwenPaw --skill visual-asset-design -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install agentscope-ai/QwenPaw visual-asset-design --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/agentscope-ai/QwenPaw.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design .gemini/skills/visual-asset-design && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "visual-asset-design" agent skill from https://github.com/agentscope-ai/QwenPaw/tree/main/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design into .gemini/skills/visual-asset-design/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "visual-asset-design", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install agentscope-ai/QwenPaw visual-asset-designInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add agentscope-ai/QwenPaw --skill visual-asset-design -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/agentscope-ai/QwenPaw.git skills-src && mkdir -p .github/skills && cp -r skills-src/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design .github/skills/visual-asset-design && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "visual-asset-design" agent skill from https://github.com/agentscope-ai/QwenPaw/tree/main/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design into .github/skills/visual-asset-design/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "visual-asset-design", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add agentscope-ai/QwenPaw --skill visual-asset-design -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install agentscope-ai/QwenPaw visual-asset-design --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/agentscope-ai/QwenPaw.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design .opencode/skills/visual-asset-design && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "visual-asset-design" agent skill from https://github.com/agentscope-ai/QwenPaw/tree/main/plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design into .opencode/skills/visual-asset-design/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "visual-asset-design", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
visual-asset-designGuides writing and repairing image prompts for character, scene and prop reference sheets, including layout, camera language and continuity across variants.
The skill compiles creative briefs for characters, scenes and props into executable image-generation prompts. Your requirements outrank the image model's requirements, which outrank the skill's defaults, and cost rules such as one image per variant are left to the main agent's system prompt. Reference sheets follow a hierarchy: a large primary view plus smaller separate views, with no equal-sized grids and no repeated or overlapping panels.
Characters get a cinematic art identity board with a hero full-body view, silhouette, expression and detail studies and a minimal ID block. Scenes get a multi-view environment board kept spatially consistent from every camera position, so left and right placements are checked against fixed landmarks. Sequence keyframe sheets use square grids with equal rows and columns so each cell matches the target aspect ratio. It also covers ways to avoid false positives from image moderation, entity continuity, and deriving variants from a main reference, such as day and sunset versions of one place. Video prompts belong to professional-media-prompts.
6 steps, taken from the step headings in SKILL.md.
Read from SKILL.md and the folder at commit d7a0f50. It shows what the files ask for, not the result of running them.
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
No scripts in the folder and no shell commands in SKILL.md.
From the folder's file list and the shell code blocks in SKILL.md.
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Visual Asset Design loads about 2.8k tokens when it runs. Until then it costs about 57 tokens; SKILL.md has 457 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
The full file from agentscope-ai/QwenPaw at commit d7a0f50, republished under its Apache-2.0 licence (© agentscope-ai). 457 words, ~2,816 tokens.
.claude/skills/visual-asset-design/SKILL.md (or your agent's skills folder).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"。
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:
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。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。
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:
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。适用场景(不限于剧情):
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:
默认在一张图中以较大的主体三分之四视图或最有辨识度的角度建立完整外观,再用独立、较小的正/侧/背面、结构、功能部位和材质细节补充信息。视图大小有主次、轮廓完整、间距清楚,不默认等大宫格。所有视图使用同一道具、同一状态、相同尺度与材质;磨损、湿润、打开或其他状态变化按主图衍生规则制作。用户要求单图或指定布局时遵循用户要求。
道具研究图默认只展示道具及其必要细节,不额外画主角作为装饰或比例参照。需要表达尺度时优先使用尺寸关系文字描述或无身份的中性参照;确实需要同框角色时,必须引用该角色当前选定的身份图,并在 prompt 中明确遵循最新外观,不能从旧图或旧描述补画角色。场景空镜、道具图与角色图各自承担清晰的参考职责。
| 用途 | 默认布局 | 用户可调整 |
|---|---|---|
| 角色首次建立身份 | 电影感艺术身份板(不对称层级) | 用户可要求标准设定表、单图或指定视角 |
| 道具首次建立身份 | 多视角道具研究板(大型主体 + 角度/结构/材质细节) | 用户可要求单图、指定视角或布局 |
| 场景/环境建立 | 多视角环境研究板(大型全景 + 空间角度与局部细节) | 用户可要求单图、指定视角或布局 |
| 主体在一段内容中的关键瞬间 | 序列关键帧参考图(多宫格时间线) | 用户可要求单图或指定宫格数 |
优先级:用户明确要求 > 默认推荐。
为 VisualEntity 生成参考图时,Prompt 按以下维度组织;不要只写“主体/姿态/环境/风格”四句概括:
角色首次建立身份时默认使用电影感艺术身份板;角色在内容中的关键瞬间使用序列关键帧;场景锚点使用多角度空镜或单图全景;道具使用英雄视图与多角度细节。用户有明确要求时按用户要求执行。
角色外观、服装、配饰或风格改变后,先检查受影响片段实际绑定的整组参考图,再恢复分镜与视频生成。检查关联 Variant 的 prompt 和选中图片本身,尤其是道具图中的比例人物、场景中的背景角色、阵容图和复合设计图;不能仅凭实体类型或文件名认定它们与修改无关。
由 Agent 根据当前用户反馈决定哪些图片冲突、哪些可以复用,不使用固定关键词规则,也不默认重做全部资产。确认存在旧形象时,更新该资产 prompt 与引用后生成一致版本,或选择已有且满足用途的纯道具/空镜图;在同一次返修提交中同步处理受影响目标的旧选择和下游显式参考列表。新图未就绪时保持相关生成 prompt 未就绪,避免分镜消费新角色图与旧角色参照的混合输入。重新绑定后按最终参考顺序更新 [Image N],实际检查过图像才可声明一致性已确认。
生成参考图时,根据主体类型和叙事需求选择合适的构图方式:
优先级:用户明确要求 > 默认推荐。
图片模型的输出审核(如 DashScope 的 green-net)拒绝是确定性的:同样的 prompt 每次都会被拒,且依赖该资产的全部分镜与视频节点都会被卡住。所以写 prompt 时就要避开常见误判词族,而不是等失败后再改。
根据项目风格选择合适的质感表现:
同一实体下的多个 Variant 必须:
同一主体只改变状态、服装、时间、天气或材质表现时,应基于主图衍生,避免每个 Variant 独立从文字重新设计。此规则同时适用于角色、场景、道具,由主 Agent 根据用户意图选择基准并维护引用。
canonical_variant_id。用户上传参考图时,以符合用户要求的上传图作为主图生成依据;有可复用且已确认的主图时直接复用。derived_from_variant_id,在 requirements 中写清保留项和变化项。该字段只是关系记录,不能替代真实参考图绑定。主图未生成、未通过当前模式要求的审阅,或正在返修时,衍生 prompt 保持为空;先推进其他不依赖它的工作,不轮询等待,也不发布缺参考的衍生生成请求。reference_artifact_version_ids,并在同一次提交中写入完整 prompt。生成前确认该版本仍是符合当前用户要求的主图;不虚构未来版本 ID,不引用已撤销的主图,也不假定运行时会根据衍生字段自动补图。例如:先完成干燥的 Q 版噜噜身份板;湿噜噜引用这张主图,保留大头小身、脸型、毛色和配饰,只增加湿润反光、水珠及必要的毛发变化。先完成白天的庭院水池环境板;夕阳版引用白天主图,保持水池形状、地标和路径,只改变时间、光色与阴影。道具从完好到磨损、闭合到打开时同样以原主图为基础,明确允许改变的结构或表面状态。
主图因用户反馈被替换后,检查现有衍生图和下游引用是否仍符合新要求,由 Agent 决定复用或修订;不要继续用被撤销的图恢复旧形象,也不要无条件重做全部资产。
source_asset_version_id 写入 reference_asset_version_ids;衍生图的主图 ArtifactVersion 写入 reference_artifact_version_ids。两类字段的版本必须真实存在,不能互相混写。visual.entities 中的稳定 ID 维护,不在 Element 中重复声明角色视觉细节。reference_asset_version_ids、reference_artifact_version_ids,随后才是未被这些参考覆盖的角色主图。按这个顺序编写 [Image N] 职责,不把 character_refs 的站位顺序当成图片编号顺序。character_refs、截断参考列表或虚构版本 ID 绕过校验。空间上相邻或连通的场景属于同一个空间群:同一建筑内的教室与教室外的走廊、同一住所的卧室/家门口/楼道电梯口、同一街区的相邻店面、同一庭院的不同角落。空间群内的场景共享建筑语言、材质体系、色调与光照逻辑,各自只从文字或角色图出发会产生风格漂移。
reference_artifact_version_ids,形式为风格锚点 visual:<entityId>:<variantId>(与角色锚点、素材参考并列)。工作图会让这些场景自动等待基准图生成并获准后,携带基准图作为参考图再生成——不需要也不应该把相邻场景的 prompt 留空等待。[Image N] 职责中写清基准图提供的是共享的建筑语言、材质体系、色调光照与连接结构(同一扇门、同一段墙面、同一排窗、同款地面),不是构图模板——机位、景别与空间内容仍按新场景自身的需要写。不同实体类型在 Prompt 中的侧重点:
clear spatial labels 与 no text;只有用户明确需要导演审阅图时才允许短标签,并把这些标签列为文字排除项的显式例外。无文字模式也不得用字母 A/B/C、地名或图例文字暗中代替标签。以下创作品味准则源自 vendored review_rubrics(Qwen-MM-Plugins video-edit skill),应当遵循(建议性准则,不是门禁):
© agentscope-ai, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Just SKILL.md in plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design of agentscope-ai/QwenPaw.
Open the folder on GitHubat commit d7a0f50
Visual Asset Design next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Visual Asset Design this skillagentscope-ai/QwenPaw | 36k | — | ~2.8k | Automated safety check: Pass | Apache-2.0 | |
| Canghe Comicfreestylefly/canghe-skills | 461 | 8 repos | ~3.2k | Automated safety check: Pass | None | |
| Xhs ImagenNimaChu/xhs-imagen | 120 | — | ~2.3k | Automated safety check: Pass | MIT | |
| Hbg Life SimulationMr-funny/hbg-life-simulation | 140 | — | ~5.8k | Automated safety check: Pass | MIT | |
| Knowledge Comic CreatorJimLiu/baoyu-skills | 26k | 2 repos | ~5.5k | Automated safety check: Pass | MIT | |
| Codex Image Servicezrt-ai-lab/opencode-skills | 287 | — | ~2k | Automated safety check: Pass | None |
freestylefly/canghe-skills
Knowledge comic creator supporting multiple art styles and tones.
NimaChu/xhs-imagen
Create complete Xiaohongshu image-and-text posts from a topic, article, document, or rough idea, including research, fact checking, post copy, pagination, storyboards, a 3:4 cover, 9:16 content…
Mr-funny/hbg-life-simulation
Create Chinese HBG “模拟人生 / 人生副本” narrative videos with a consistent comic IP, a rapid multi-life opening, continuous natural-speed narration, synchronized short captions, dense static manga…
JimLiu/baoyu-skills
Writes original educational comics with detailed panel layouts in a chosen art style and tone, generating the images in batches through whichever backend is available.
zrt-ai-lab/opencode-skills
A skill your agent uses when Codex needs native image generation, editing, or engineered multi-image composition, especially sets, material kits, grids, storyboards, carousels, presentations…
fal-ai-community/skills
Builds consistent characters, reference and expression sheets, outfit variations and character-to-video shots with the genmedia CLI while keeping identity stable.
agentscope-ai/QwenPaw
Guidance for writing and testing Terraform and OpenTofu code: module structure, naming, test approaches, CI/CD workflows, state handling and security scanning.
agentscope-ai/QwenPaw
Turns reusable decisions, templates or workflows from the current conversation into a new workspace skill through plan, approval, draft, validation and publication.
agentscope-ai/QwenPaw
Creates a focused workspace skill from the current conversation in QwenPaw, moving through a planning, approval, drafting, validation and publishing script pipeline.
agentscope-ai/QwenPaw
Creates and manages scheduled or recurring jobs with the qwenpaw cron commands, always tied to an explicit agent ID and a confirmed target channel.
agentscope-ai/QwenPaw
Creates, reads and edits Word .docx files, including tracked changes and comments, using docx-js for new files and XML editing for existing ones.
agentscope-ai/QwenPaw
Connects, registers, and operates a personal mailbox, reading, searching, sending, and organizing, through a managed mail server for nine domains.
Categories
Guides writing and repairing image prompts for character, scene and prop reference sheets, including layout, camera language and continuity across variants. The skill compiles creative briefs for characters, scenes and props into executable image-generation prompts. Your requirements outrank the image model's requirements, which outrank the skill's defaults, and cost rules such as one image per variant are left to the main agent's system prompt.
Visual Asset Design fits situations like: designing a reference sheet for a new character, scene or prop; fixing an image prompt whose views contradict each other; deriving a sunset version of a scene from its main reference.
Run `npx skills add agentscope-ai/QwenPaw --skill visual-asset-design -a claude-code`. Or copy the skill folder (plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design in agentscope-ai/QwenPaw) into .claude/skills/visual-asset-design in your project. Claude Code loads it when a task matches its description.
Run `npx skills add agentscope-ai/QwenPaw --skill visual-asset-design -a codex`. Or copy the skill folder (plugins/apps/qwenpaw-creator/backend/skills/visual-asset-design in agentscope-ai/QwenPaw) into .agents/skills/visual-asset-design in your project. Codex loads it when a task matches its description.
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add agentscope-ai/QwenPaw --skill visual-asset-design -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/visual-asset-design, .gemini/skills/visual-asset-design, .github/skills/visual-asset-design and .opencode/skills/visual-asset-design in your project.
SKILL.md names no scripts, command-line tools or credentials: Visual Asset Design is instructions for the agent only. Our summary lists: QwenPaw Creator with an image generation model.
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
Visual Asset Design is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
About 2.8k tokens (SKILL.md is roughly 11k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
Skills that share tags, products or a category with Visual Asset Design: Canghe Comic (freestylefly/canghe-skills, 461 stars), Xhs Imagen (NimaChu/xhs-imagen, 120 stars), Hbg Life Simulation (Mr-funny/hbg-life-simulation, 140 stars) and Knowledge Comic Creator (JimLiu/baoyu-skills, 26k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
agentscope-ai (a GitHub organization) maintains it in agentscope-ai/QwenPaw, which has 35,521 GitHub stars. The repository holds 20 skills in this directory. The repository was last updated on October 9, 2026.
Source: agentscope-ai/QwenPaw on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.