MCP Server Builder
anthropics/skills
Guides the design and implementation of Model Context Protocol servers in TypeScript or Python, from tool naming and error messages to evaluation.
使用 PuppetLoom 让外部 Agent 通过 CLI 从分层角色 PSD 创建或改善可动 2D 角色,可按整模或指定部位完成分析、网格与运动制作、自检、返修和准确 revision 的证据闭环;也用于检查关键形、权重、透视、次级运动与校准结果,通过 CLI 依次演示编辑器和角色窗口的真实功能,以及直接导出 Cubism 工程和运行时文件或使用 Editor 桥接。Use when an…
$ npx skills add CheshireMew/PuppetLoom --skill live2d-puppet -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install CheshireMew/PuppetLoom live2d-puppet --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/CheshireMew/PuppetLoom.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/live2d-puppet .claude/skills/live2d-puppet && 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 "live2d-puppet" agent skill from https://github.com/CheshireMew/PuppetLoom/tree/main/skills/live2d-puppet into .claude/skills/live2d-puppet/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "live2d-puppet", 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/CheshireMew/PuppetLoom/tree/main/skills/live2d-puppetType 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 CheshireMew/PuppetLoom --skill live2d-puppet -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install CheshireMew/PuppetLoom live2d-puppet --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/CheshireMew/PuppetLoom.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/live2d-puppet .agents/skills/live2d-puppet && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "live2d-puppet" agent skill from https://github.com/CheshireMew/PuppetLoom/tree/main/skills/live2d-puppet into .agents/skills/live2d-puppet/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "live2d-puppet", 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 CheshireMew/PuppetLoom --skill live2d-puppet -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install CheshireMew/PuppetLoom live2d-puppet --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/CheshireMew/PuppetLoom.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/live2d-puppet .cursor/skills/live2d-puppet && 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 "live2d-puppet" agent skill from https://github.com/CheshireMew/PuppetLoom/tree/main/skills/live2d-puppet into .cursor/skills/live2d-puppet/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "live2d-puppet", 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/CheshireMew/PuppetLoom.git --path skills/live2d-puppet--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 CheshireMew/PuppetLoom --skill live2d-puppet -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install CheshireMew/PuppetLoom live2d-puppet --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/CheshireMew/PuppetLoom.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/live2d-puppet .gemini/skills/live2d-puppet && 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 "live2d-puppet" agent skill from https://github.com/CheshireMew/PuppetLoom/tree/main/skills/live2d-puppet into .gemini/skills/live2d-puppet/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "live2d-puppet", 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 CheshireMew/PuppetLoom live2d-puppetInstalls 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 CheshireMew/PuppetLoom --skill live2d-puppet -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/CheshireMew/PuppetLoom.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/live2d-puppet .github/skills/live2d-puppet && 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 "live2d-puppet" agent skill from https://github.com/CheshireMew/PuppetLoom/tree/main/skills/live2d-puppet into .github/skills/live2d-puppet/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "live2d-puppet", 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 CheshireMew/PuppetLoom --skill live2d-puppet -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install CheshireMew/PuppetLoom live2d-puppet --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/CheshireMew/PuppetLoom.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/live2d-puppet .opencode/skills/live2d-puppet && 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 "live2d-puppet" agent skill from https://github.com/CheshireMew/PuppetLoom/tree/main/skills/live2d-puppet into .opencode/skills/live2d-puppet/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "live2d-puppet", 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.
live2d-puppet使用 PuppetLoom 让外部 Agent 通过 CLI 从分层角色 PSD 创建或改善可动 2D 角色,可按整模或指定部位完成分析、网格与运动制作、自检、返修和准确 revision 的证据闭环;也用于检查关键形、权重、透视、次级运动与校准结果,通过 CLI 依次演示编辑器和角色窗口的真实功能,以及直接导出 Cubism 工程和运行时文件或使用 Editor 桥接。Use when an…
Live2d Puppet is an agent skill from CheshireMew/PuppetLoom. 使用 PuppetLoom 让外部 Agent 通过 CLI 从分层角色 PSD 创建或改善可动 2D 角色,可按整模或指定部位完成分析、网格与运动制作、自检、返修和准确 revision 的证据闭环;也用于检查关键形、权重、透视、次级运动与校准结果,通过 CLI 依次演示编辑器和角色窗口的真实功能,以及直接导出 Cubism 工程和运行时文件或使用 Editor 桥接。Use when an external agent such as Codex should create or improve a PuppetLoom character through the CLI, review exact-revision visual evidence, demonstrate the real editor and runtime under CLI control, or bridge a verified project to official Cubism runtime files. 不用于在桌面应用里内嵌 Agent 对话框、让用户默认承担逐点调网格、纯视频生成或没有相关角色/模型的普通图像编辑。
Its SKILL.md is about 2.7k tokens, which your agent loads only when the skill is triggered. The skill folder holds 37 other files, including scripts, reference files and assets (for example `agents/openai.yaml`, `references/agent-review-and-repair.md` and `references/asset-and-geometry-workflow.md`).
It works with TypeScript. The licence is AGPL-3.0.
Read from SKILL.md and the folder at commit 772b32f. 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.
Ships 1 file in scripts/, which the agent can run.
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.
Live2d Puppet loads about 2.7k tokens when it runs, and up to ~33k if it reads all its reference files. Until then it costs about 132 tokens; SKILL.md has 496 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); the scripts in this folder are not scanned.
The full file from CheshireMew/PuppetLoom at commit 772b32f, republished under its AGPL-3.0 licence (© CheshireMew). 496 words, ~2,672 tokens.
.claude/skills/live2d-puppet/SKILL.md (or your agent's skills folder). This skill also uses 33 other files; get the full folder from GitHub.把角色原图、See-through 分层 PSD 和自然语言目标交给 Codex 一类外部 Agent。Agent 通过 PuppetLoom CLI 自动创建项目,再按整模或指定部位完成结构分析、网格与关键形制作、权重和物理配置、自检、局部返修及准确 revision 的视觉证据。用户主要审查角色实际效果,并用“幅度小一点”“滞后太慢”“太弹”这类结果反馈继续调整;逐点编辑只是备用能力。桌面应用是查看、播放和人工兜底工具,不承载 Agent 编排或聊天入口。默认交付仍是普通 PuppetLoom 项目。用户明确需要 Cubism 时,使用 cubism export 输出 .cmo3/.moc3 和配套文件;已有 Editor 工程的同步仍可使用官方 API 桥接。直接导出与 API 同步的限制必须分别报告。第一次结果以“合理地会动”为目标,不用十五种嘴形或未经素材支持的复杂表情;需要实时麦克风口型时,默认只使用已有闭口和一个张口状态。
先使用包装脚本运行 capabilities --json,确认 build.current、实际命令和配套 Skill 身份;不要根据软件版本号猜工具是否存在。包装脚本支持 PUPPETLOOM_ROOT 和 PUPPETLOOM_NODE;复制到仓库外的 Skill 必须明确项目根。按当前结果读取对应方法,不为一次局部修形加载全部制作流程:
| 当前任务 | 读取方法 |
|---|---|
| 从 PSD 创建角色或更新源 PSD | references/from-zero-workflow.md |
| 分部制作、检查与返修 | references/agent-review-and-repair.md;涉及形状或运动时加读 references/visual-rigging-rules.md |
| 原素材闭眼、脸缘和五官的目标修形 | references/shape-target-workflow.md;视觉标准沿用 references/visual-rigging-rules.md |
| 拆发、替换头饰等通用补件,或局部关键形修形 | references/asset-and-geometry-workflow.md;视觉判断沿用 references/visual-rigging-rules.md |
| 缺原图、See-Through 分层、源图错误 | references/source-art-and-layering.md;明确选择本地推理时再读 references/see-through-local-deployment.md |
| 录屏、编辑器和角色窗口演示 | references/runtime-demonstration.md |
| 用户接受、拒绝、手动校准或学习反馈 | references/calibration-and-learning.md |
| Cubism、moc3、model3 或官方工具兼容 | references/cubism-bridge-workflow.md |
使用 scripts/invoke_puppetloom.ps1 调用公开 CLI。当前任务已经有明确项目或仓库根目录时,测试、生成、中间、最终产物以及日志、截图、录屏和迁移暂存全部写在该根目录内;依赖、缓存和桌面应用用户数据留在机器工具目录,不冒充项目产物。普通项目先 inspect/create/verify/describe/render,再按任务和用户允许的证据形式选择 play 或 record;实际查看姿态图、4×4 次级运动图、准确 revision 的实时窗口,或获准生成的动态接触表/WebM 后才能判断结果。完整角色交付前还必须对同一个规范项目和准确 revision 运行 performance,以活动态与暂停态对照的 FPS、p95、p99、最差帧和长帧判断真实渲染是否稳定;固定 fixture、平均 FPS 或 ok: true 不能代替 valid: true。测试通过不等于视觉自然,视频也不是固定交付物。分层验收发现 See-Through 背景残片、低透明托底或可安全去除的硬矩形托底时,不把普通创建的保守 Alpha 清理冒充 PSD 已修复;先由外部 Agent 结合原图、单层和未增强合成确认异常边界,再通过同一包装脚本的 psd repair/review/finalize 生成新的 PSD、绑定视觉证据并形成不可改写的终态。Photoshop 已经运行时不得连接或复用用户会话,先让用户保存并关闭 Photoshop;自动化只能使用本次命令启动的会话,且有剩余文档时必须保留窗口可见、禁止退出。用户提供多份 PSD 或要求结合原画补件时,先建立逐部件来源清单并选定规范画布;不同分辨率 donor 只有在明确启用整画布等比例适配且宽高比一致时才能进入 Photoshop 配方。原 PSD 永不覆盖,未完成逐层细节与 Alpha 复核、未形成视觉终态或被拒绝的输出不得进入创建;详细边界和命令见 references/source-art-and-layering.md 与 references/from-zero-workflow.md。
角色闭眼与头脸造型先按 references/shape-target-workflow.md 标记实际结构,使用 fit-landmarks / curve-warp 和 author preview/apply 制作可核对的目标;自动起稿不是最终形状。其他分部制作的正式入口是外部 Agent 先看基线,再用 agent specification 取得当前 revision 的结构化模板,填写 goal/anatomy/parts/intent/rationale,最后运行 agent plan/apply --spec <json>。anatomy 是本角色的结构事实,不是通用风格参数:头脸和嘴保留实际眼线、嘴线与中轴,前后发写实际发束及根部,左右耳分别写实际图层、轴心和全顶点固定/释放权重,头饰与上衣也写各自连接轴心和全顶点权重;头饰还必须明确选择 headwearPerspective: "crown" 或 null,不得根据长宽比或旧角色形状猜。每个部位都必须填写基于准确 revision 画面的非空 rationale;原样模板、占位理由、空壳 anatomy、未覆盖实际图层或缺少理由会被 CLI 拒绝,不能用安全默认值冒充视觉判断。parts 可覆盖头脸、眼睛、嘴、前发、后发、呆毛、耳朵、头饰、身体、上衣、裙摆、尾巴和配饰。先运行只读 agent plan,核对 inputMode: structured-specification、baseRevision、草稿接管、目标图层、anatomy 覆盖、各部位状态、检查、返修、素材请求和 blockers;计划符合目标后再运行 agent apply。自然语言 --instruction/--scope 只保留只读计划兼容,不能正式 apply 或创建 revision:自然语言理解、看图判断和返修决策由本 Skill 中的外部 Agent 承担,软件只执行和验证明确规格。整模执行按确定顺序处理所有存在的部位,每个部位单独形成可恢复 revision,最后返回整模 verification、总状态和报告路径。项目没有相应图层时报告 not-present,现有结构无法支持目标表情且确实缺少素材时报告 needs-assets,不能伪造图层或把它们说成已经制作完成。
“完成整模”“检查遗漏”或“继续把现有角色做好”不等于立即运行整模 apply。已有 revision 和接受记录的成熟项目先运行 history/verify/describe,查看当前准确 revision 的整模及所有实际存在部位,并列出必须保留的已接受结果;只为确实观察到的缺陷生成 selected 或单部位规格。若没有待修缺陷,不执行 apply,不制造无意义 revision。CLI 的整模 apply 不是整体事务:已完成部位会逐个提交,后续部位变成 blocked 不会自动撤销前面的 revision;返回后必须核对每个部位的实际 from/to revision、session 和报告,不能只看总状态。
agent front-hair plan/apply 与 agent secondary plan/apply 仍是精确控制和兼容入口;正式整模或分部制作优先使用顶层结构化规格入口。标准表情和动作库使用 actions plan/apply。参数结构使用 author inspect/apply;局部关键形先用 author geometry 分页定位,再以选区和 transform-keyform 修形,不让模型猜整层顶点数组。轴心和权重仍用 describe --layer <id> 与稀疏 calibrate。固定连接与自由端的判断由 references/visual-rigging-rules.md 负责;补件归位和试装由 references/asset-and-geometry-workflow.md 负责。edit 只在用户明确想手调,或公开 CLI 确实无法表达局部修正时打开,不用扩建手工编辑器代替外部 Agent 制作。
范围必须服从用户目标:用户要求整模时审查整模,真正写入只覆盖已证明需要修复的部位;用户只点名一个部位时只使用对应部位 ID,不擅自扩展。用户要求调整“整个图层顺序”时,范围是当前角色全部可见遮挡关系,不会被最后指出的某个脖子、兜帽或配饰问题收窄。用户在任务中追加的限制立即作用于后续操作;后来的截图、批注或单项纠正默认只修正点名项,除非用户明确替换目标,不得把此前同一任务的完整范围收窄成最后一个问题。用户说“只使用现有素材”时,不生成、下载、增强或改写 PSD/纹理,needs-assets 只报告缺口;用户说“不要视频”时不运行 record,改看 render --project <directory> --output <new-directory> --revision <n> --size 1080 --focus <part>、4×4 运动表和 play --project <directory> --revision <n>。已有接受结果必须保留可恢复 revision,并作为未点名区域的保护基线;用户明确把某个项目作为只读对照时,另建获准的工作副本并核对对照项目未变。现有项目只缺多房束、侧脸深度或躯干体积等后来增加的字段时,先使用 extensions plan/apply 在当前项目追加可恢复 revision,不得重新创建项目;软件升级、CLI 新能力或项目格式新增字段不等于源 PSD 变化。源 PSD 确实变化时,用 migrate 在规范项目内部的新暂存目录建立候选,完整复核后归档旧运行状态并把候选晋升回原规范路径,不能在旁边留下 Character-v2 作为第二个最新版。每次自动制作后都比较修改前后,并查看准确 revision 的连续运动:安全姿态通过只证明没有已知翻转,不能证明目标部位真的在动、根部没有脱离、原有效果没有退化。Agent 必须实际打开各部位返回的 focusComparisonSheet 和 4×4 focusMotionSheet,检查连续帧并记录仍存在的问题;focusMotionManifest 只用于定位需要放大的单帧,不得只读 JSON 后宣布完成。Cubism 请求使用同一包装脚本的 cubism 命令族:直接导出走 cubism export;已有 Editor 工程的检查、预览或同步才走原 API 桥接计划。
输入获取也由外部 Agent 编排。原画生成后先展示并等待用户确认,再固定列出三种路线:推荐用户在指定 ModelScope 网页以 1024 自行上传并下载 PSD;用户选择偷懒选项后由 Agent 自动上传这张已确认原画;在线入口不可用时才推荐本地部署。选择 Agent 代传本身就是对本次具体上传的授权,不再重复询问;没有选择就不得上传。本地复核用户带回的 PSD 不等于本地部署 See-Through。本地安装、硬件门槛、已知问题和优化路线以 references/see-through-local-deployment.md 为真源;项目画风参考、唯一生图交接、1024 默认分辨率、单主体素材边界、上传副本和三档视觉验收以 references/source-art-and-layering.md 为真源。用户只要求“生成 PSD”“测试分层”或“看看 See-Through 是否可用”时,范围严格停在原图、实际上传副本、PSD、真实可见图层重组和已定稿的视觉结论;不得创建 PuppetLoom 项目、编写绑定规格或进入 agent plan/apply,除非用户随后明确要求继续制作。测试输入必须是真正打开检查过的正常日系二次元角色图,不能用抽象图或几何测试卡。实际生图继续由外部图像工具承担,不把模型、提示词引擎或 Agent 运行时塞进 PuppetLoom。参考视频和获准生成的姿态参考图只作观察证据,不进入 PSD、纹理或项目素材;最终运动范围仍服从原素材能支撑的透视。
用户要求创建完整角色时,先判断现有眼白、虹膜和睫毛能否支持几何眨眼;缺闭眼贴图不等于缺眨眼能力。只有几何闭合不能保持角色眼形、眼睛烘焙在脸部或目标需要特定手绘造型时,才补闭眼素材。一个张口等确实缺少的必要素材由外部 Agent 生成、检查并接入,不再逐项索取授权。已有闭嘴就是 mouthOpen=0,不得重画;微张嘴和音素嘴形只有用户明确需要对应表情或触发方式时才增加,不作为麦克风口型的默认状态。用户明确说“只使用现有素材”“不要生图”时才停止补素材并报告 needs-assets。生成闭眼前先确认睁眼结构是否独立;睁眼已经烘焙在脸部时,闭眼素材必须完整遮住原眼或先修 PSD,不能只叠两条眼线。闭眼和嘴形必须继承原画的线稿、睫毛体量、眼角、口腔配色、阴影和抗锯齿,不能用一条弧线或粗糙符号糊弄;完整规则见 references/visual-rigging-rules.md。
一次角色校准只属于当前项目。候选 revision 先做足以安全展示的定向检查并交给用户看;用户接受后用 evidence --project <directory> --session <id> --status accepted 记录,再运行获准范围内的完整回归或推送。任何相关 session 仍是 unreviewed 时只能称为候选,不能报告“完成”“已修好”或把它当成以后角色的模板;用户拒绝时,记录被否定的 session 和具体缺陷;需要撤回时只恢复受影响范围,不撤销其他有效工作。用户同时给出新方向或已授权继续修复时,重新看图并返修该项,不因拒绝旧候选而停止任务,也不把尚未解决的问题改记为接受。用户给出同一部位的新方向时,优先把结果反馈交给现有 Agent 或高层 authoring 入口,不要求用户指出顶点编号。多个独立项目反复出现同一错误,是修改 PuppetLoom 通用算法的充分证据但不是唯一入口;一个真实项目若已用代码与数据证明是可复用的软件缺陷,也可在增加通用 fixture、真实项目回归和非退化证据后修软件。角色专属顶点、轴心和权重仍留在项目校准中,不能升级成其它角色的默认几何。用户明确要求同步优化本 Skill 时,把已经验证的制作方法、入口变化和错误判断一并修正,检查主说明与各引用文件是否矛盾;不自动调用其它 Skill,也不对已获授权的修改重复询问。没有修改授权时只记录建议。修改前后运行 scripts/file_budget.py。
references/from-zero-workflow.md:输入边界、完整 CLI 顺序、Agent-first 修改优先级、PSD 迁移、静态/动态证据检查和桌面交接。references/source-art-and-layering.md:从角色需求取得可绑定原图、官方在线 See-Through 默认交接、获准自动 API 与最后可选的本地部署边界、单条生图指令和 PSD 重组验收。assets/blue-whale-maid-reference/:项目长期参考素材;鲸鱼娘原画保留作可选观察参考,其可复用的正向画风已经写入原图流程,只有用户明确要求使用整张参考图时才进入生图上下文;分层 PSD、转头图和表情图只供分层、绑定和补表情时查看。references/see-through-local-deployment.md:本地 See-Through 的启用条件、C/D/E 盘环境检查、硬件与分辨率门槛、已知问题和后续优化顺序。references/agent-review-and-repair.md:外部 Agent 与软件的职责边界、结构化制作规格、看图—返修循环、缺陷路由和用户验收。references/visual-rigging-rules.md:从本项目历史中保留下来的结构、透视和次级运动判断。references/shape-target-workflow.md:角色结构标记、眼睑分区、脸形目标、两种目标变换、只读预览与按缺陷返修。references/asset-and-geometry-workflow.md:通用补件的参考、配准、试装与恢复,自然拆发和局部关键形的区域操作。references/runtime-demonstration.md:编辑器优先、角色窗口在后、公共 runtime CLI 控制、只读演示和可靠窗口保活。references/calibration-and-learning.md:用户校准如何进入项目、软件或 Skill,及其证据门槛。references/cubism-bridge-workflow.md:官方格式边界、参数映射、Editor 版本与授权、严格同步、侧车生成、最终目录和视觉验收。scripts/invoke_puppetloom.ps1:定位 PuppetLoom、必要时构建并原样转发包括 psd repair/review/finalize 在内的 CLI 命令和退出码。scripts/acquire_layered_psd.ps1、scripts/acquire_layered_psd.py 与 scripts/finalize_psd_review.py:通过 ModelScope See-Through 的公开 Gradio API 上传已获准原图,或在本地复核用户带回的原图和 PSD;生成原图与实际上传副本、归一化原图、带名称预览、真实可见图层重组、对照图、检查记录和一次传输重试证据,并确定性定稿三档视觉结论。它们不启动或安装本地 See-Through 推理服务。requirements-layering.txt:PSD 重组与对照图所需、已在 D:\Tools\Python310 验证的 Pillow 和 psd-tools 版本。scripts/demo_puppetloom.ps1 与 scripts/demo_puppetloom.mjs:通过 CLI 驱动真实 Electron 完成可录制的只读演示,并按需保持窗口。scripts/file_budget.py:Skill 自我修改前后的本地活动文本预算门槛。PSD-only 任务返回同一个自包含运行目录,至少保留原图、实际上传副本、PSD、图层预览、inspect、真实图层重组对照、visual-review.json 和 result.json。Agent 必须实际看图,把结论记录为 accepted、accepted-with-repairs 或 rejected,再用 scripts/acquire_layered_psd.ps1 -FinalizeReview 校验并同步唯一结论;结构检查和数值差异不能代替视觉判断。accepted-with-repairs 只容纳不改变主要语义、轮廓和遮挡结构的小问题,并在用户明确接受携带这些问题时才进入创建;真正阻止可靠创建的问题进入 blockingIssues 并拒绝候选。完成 PSD-only 范围后立即停止,不返回项目目录或绑定规格。
创建任务返回项目目录、绑定等级、禁用功能和视觉证据路径;整模或分部任务返回 scope、自然语言目标、from/to revision、各部位的 awaiting-visual-review/not-present/needs-assets/blocked 状态、实际操作、前后对比、所采用的连续证据、最终 verification、总 blockers 和任务报告路径。用户问“这轮最新版”“都解决了吗”或“看看解决了哪些”时,先从完整任务、批注和纠正中重建需求清单,再对每项已批准要求沿同一条完成链核对:软件与公开 CLI 确实具备能力,普通默认路径或用户指定入口实际启用了它,规范项目的准确 revision 已保存对应数据,匹配该 revision 的视觉证据已经查看,依赖观感的结果已经交给用户判断,并且素材或官方工具边界已如实报告。完成链中任何一环没有成立,就只报告已经到达的层级和剩余工作,不能宣称整项完成;隐藏的可选开关、测试副本、产物目录和单独通过的 fixture 都不能代替正式用户路径与规范项目。已有项目只保留一个规范项目目录,以内部 revision、session 和证据历史作为版本真源,不复制 project-rN 目录冒充版本管理。交付前检查本任务的测试、生成、中间和最终文件都位于规范项目或当前仓库根目录内,再用桌面端 edit 或 play 实际打开该规范路径,并确认“最近项目”记录的是同一个绝对路径;只完成 CLI create 或打开了暂存副本都不算完成。只有 verification.valid、完整角色准确 revision 的 performance.valid、实际姿态与每个准确 revision 的允许证据都看过、报告中的项目指纹和窗口比例匹配、目标部位产生以源图像像素计可见而合理的运动、未点名区域和已接受结果没有明显退化、用户要求的角色窗口或编辑器链能够打开,才算普通项目结果在客观范围内成立。依赖观感的幅度、延迟和回弹在用户判断前只是待确认结果,不能因为自检全绿就宣称完成。
Cubism 直接导出任务还要返回源 revision、指纹、工程与运行时版本、参数 ID 映射、CMO3/model3 路径,以及 Core 和目标 Editor 的实际验证结果。cubism export 的 awaiting-visual-review 表示尚待看图;必须对照实际导出图集、UV、九向、闭眼及中间帧,在 Editor 打开并保存后才能报告对应范围通过。cubism verify 只证明目录结构和引用。API 桥接另行报告 Editor/API 状态、兼容计划、strictReady 和 blocking/warning;严格同步或用户已接受的 partial 结果仍需 Cubism Viewer 复核。需要用户判断时展示前后画面并说明具体变化,不用“应该好了”代替视觉证据。
© CheshireMew, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
SKILL.md and 33 other files (scripts, references, assets) in skills/live2d-puppet of CheshireMew/PuppetLoom.
Open the folder on GitHubat commit 772b32f
Live2d Puppet 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 |
|---|---|---|---|---|---|---|
| Live2d Puppet this skillCheshireMew/PuppetLoom | 245 | — | ~2.7k | Automated safety check: Pass | AGPL-3.0 | |
| MCP Server Builderanthropics/skills | 180k | 63 repos | ~2.3k | Automated safety check: Pass | Apache-2.0 | |
| Web Artifacts Builderanthropics/skills | 180k | 40 repos | ~769 | Automated safety check: Pass | Apache-2.0 | |
| Typescript Advanced Typesrolling-scopes/rsschool-app | 10k | 25 repos | ~4.2k | Automated safety check: Pass | MPL-2.0 | |
| Code Review ChecklistshareAI-lab/learn-claude-code | 78k | 4 repos | ~1.1k | Automated safety check: Pass | MIT | |
| MCP Server BuildershareAI-lab/learn-claude-code | 78k | 4 repos | ~1.2k | Automated safety check: Pass | MIT |
anthropics/skills
Guides the design and implementation of Model Context Protocol servers in TypeScript or Python, from tool naming and error messages to evaluation.
anthropics/skills
Builds multi-component claude.ai HTML artifacts as a small React, TypeScript and Tailwind project, then bundles it into one shareable HTML file.
rolling-scopes/rsschool-app
Master TypeScript's advanced type system including generics, conditional types, mapped types, template literals, and utility types for building type-safe applications.
shareAI-lab/learn-claude-code
Reviews code against a five-part checklist covering security, correctness, performance, maintainability and testing, and reports findings in a fixed format.
shareAI-lab/learn-claude-code
Walks through building MCP servers in Python or TypeScript that expose tools, resources and prompts to Claude, with templates, registration and testing.
nrwl/nx
Import, merge, or combine repositories into an Nx workspace using nx import.
Works with
使用 PuppetLoom 让外部 Agent 通过 CLI 从分层角色 PSD 创建或改善可动 2D 角色,可按整模或指定部位完成分析、网格与运动制作、自检、返修和准确 revision 的证据闭环;也用于检查关键形、权重、透视、次级运动与校准结果,通过 CLI 依次演示编辑器和角色窗口的真实功能,以及直接导出 Cubism 工程和运行时文件或使用 Editor 桥接。Use when an…. Live2d Puppet is an agent skill from CheshireMew/PuppetLoom. 使用 PuppetLoom 让外部 Agent 通过 CLI 从分层角色 PSD 创建或改善可动 2D 角色,可按整模或指定部位完成分析、网格与运动制作、自检、返修和准确 revision 的证据闭环;也用于检查关键形、权重、透视、次级运动与校准结果,通过 CLI 依次演示编辑器和角色窗口的真实功能,以及直接导出 Cubism 工程和运行时文件或使用 Editor 桥接。Use when an external agent such as Codex should create or improve a PuppetLoom character through the CLI, review exact-revision visual evidence, demonstrate the real editor and runtime under CLI control, or bridge a verified project to official Cubism runtime files.
Live2d Puppet fits situations like: an external agent such as Codex should create; improve a PuppetLoom character through the CLI; review exact-revision visual evidence; demonstrate the real editor and runtime under CLI control.
Run `npx skills add CheshireMew/PuppetLoom --skill live2d-puppet -a claude-code`. Or copy the skill folder (skills/live2d-puppet in CheshireMew/PuppetLoom) into .claude/skills/live2d-puppet in your project. Claude Code loads it when a task matches its description.
Run `npx skills add CheshireMew/PuppetLoom --skill live2d-puppet -a codex`. Or copy the skill folder (skills/live2d-puppet in CheshireMew/PuppetLoom) into .agents/skills/live2d-puppet 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 CheshireMew/PuppetLoom --skill live2d-puppet -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/live2d-puppet, .gemini/skills/live2d-puppet, .github/skills/live2d-puppet and .opencode/skills/live2d-puppet in your project.
SKILL.md names no scripts, command-line tools or credentials: Live2d Puppet is instructions for the agent only.
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. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.
Live2d Puppet is published under the AGPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
About 2.7k 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. Its references folder adds about 30k tokens, read only when the agent opens those files.
Skills that share tags, products or a category with Live2d Puppet: MCP Server Builder (anthropics/skills, 180k stars), Web Artifacts Builder (anthropics/skills, 180k stars), Typescript Advanced Types (rolling-scopes/rsschool-app, 10k stars) and Code Review Checklist (shareAI-lab/learn-claude-code, 78k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
CheshireMew (a GitHub user) maintains it in CheshireMew/PuppetLoom, which has 245 GitHub stars. The repository was last updated on October 5, 2026.
Source: CheshireMew/PuppetLoom on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.