[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77P6N2ESFAQHT2WHDMHS6RG":3,"skill-same:01M77P6N2ESFAQHT2WHDMHS6RG":160},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":15,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":5,"forksCount":13,"repositoryHtmlUrl":26,"repositoryHomepageUrl":6,"ownerHtmlUrl":27,"ownerBlogUrl":6,"ownerDescription":28,"updatedAt":29,"install":30,"resources":43},"01M77P6N2ESFAQHT2WHDMHS6RG","creating-skill-pro",null,"设计、改造、融合和验证 Agent skills，在需要梳理策略框架、重构 SKILL.md 结构、规划 `references\u002F` 与 `scripts\u002F` 分层、建立评测、优化 description 或排查触发问题时使用。适用于“帮我做一个 skill”、“重写或合并这个 skill”、“给 skill 做评测”、“为什么这个 skill 没触发”等请求。","dulltackle\u002Fkangkang-skills","dulltackle","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F45963660?v=4","01M77P69WBNH9A8BYS15THR64T","01M77P69X11JVVHASHEFC2107Q",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: creating-skill-pro\ndescription: 设计、改造、融合和验证 Agent skills，在需要梳理策略框架、重构 SKILL.md 结构、规划 `references\u002F` 与 `scripts\u002F` 分层、建立评测、优化 description 或排查触发问题时使用。适用于“帮我做一个 skill”、“重写或合并这个 skill”、“给 skill 做评测”、“为什么这个 skill 没触发”等请求。\n---\n\n# Creating Skill\n\n本 skill 用于设计、改造和验证 Agent skill。默认先判断任务性质、产物和边界，再按需读取 `references\u002F` 或调用 `scripts\u002F`，不默认把所有请求推进到完整闭环。\n\n## 定位与适用任务\n\n- 从零设计新 skill，或把一套工作方式沉淀成 skill\n- 重写、分层、融合或重构已有 skill\n- 为 skill 设计评测、baseline、回归检查或结果展示\n- 优化 description、触发边界或排查为什么没触发\n- 如果只是普通文案修改、普通脚本分析或单次交付，不要误用为 skill 创建流程\n\n## 设计哲学\n\n激发模型能力上限 = 策略哲学 + 最小完备工具集 + 必要事实说明\n\n### 策略哲学\n\n- 写思考框架，不写唯一路径\n- 先判断当前任务是开放式设计问题、确定性执行问题还是验证问题\n- 只有当 workflow 能显著降低风险时才引入它；不要把 workflow 当顶层默认范式\n- 高自由度任务必须保留重规划空间，不能被固定流程吞掉\n\n### 最小完备工具集\n\n- 顶层 `SKILL.md` 只放定位、判断框架、边界和资源入口\n- `references\u002F` 提供按需细节，不默认全读\n- `scripts\u002F` 承接重复、确定性、易错的动作\n- `agents\u002F`、`assets\u002F`、`eval-viewer\u002F` 是可选能力层，不是每次都要启用\n\n### 必要事实说明\n\n- 事实说明提供推理原料，例如结构约束、环境前提、触发边界和兼容要求\n- 不要把经验偏好伪装成硬规则；事实说明负责校准判断，不负责替模型决策\n- 当事实边界不完整时，优先补信息，再决定是否进入 workflow 或校验流程\n\n## 工作原则\n\n### 先提取上下文，再补问题\n\n- 如果当前对话里已经包含 workflow、工具链、用户纠偏、输入输出样例或约束，先从现有上下文提取\n- 只有当关键信息仍缺失时，再补问题，不要让用户重复已经说过的话\n- 提问题时优先补这几类空白：触发时机、输出格式、边界、依赖资源、是否需要评测\n\n### token 是共享资源\n\n- 顶层 `SKILL.md` 只保留路由、原则、关键决策和入口说明\n- 长示例、schema、评测细节、大段领域知识放进 `references\u002F`\n- 重复、脆弱或需要确定性的逻辑优先做成 `scripts\u002F`\n\n### 沟通风格随用户熟悉度调整\n\n- 默认优先用用户意图、任务目标和产物来描述，不要上来就堆术语\n- 只有当上下文已经表明用户熟悉时，才直接使用 `eval`、`benchmark`、`assertion` 等词\n- 如果术语不可避免，先用一句中文短解释再继续，例如“评测样本（eval）”或“可验证断言（assertion）”\n- 对熟悉工程流程的用户可以更直接；对非工程背景用户要优先讲目的、步骤和交付物\n\n### 先复用，后新增\n\n- 先盘点现有 skill、脚本、参考资料、模板和历史流程\n- 能改造就不要重写，能下沉就不要堆在顶层\n- 不要为 skill 新增无关文件，如 `README.md`、`CHANGELOG.md`、安装说明\n\n### 自由度和风险匹配\n\n- 高自由度：策略建议、结构设计、边界澄清、示例组织\n- 中自由度：推荐模式明确，但允许按上下文调整\n- 低自由度：容易出错、需要严格顺序、必须稳定复现的流程\n- 高自由度任务禁止被固定流程吞掉；只有风险和收益都明确时才把它压成 workflow\n\n### description 决定是否会被用到\n\n`description` 必须同时说明：\n\n- skill 做什么\n- 什么时候必须使用\n- 用户可能怎样表达这个需求\n\n描述要优先写用户意图和使用时机，再补常见说法、相关文件类型和相邻边界。\n\n## 模型惯性与边界\n\n- 模型容易把开放式任务写成固定路径，即使当前信息不足\n- 模型容易在旧 skill 上持续打补丁，而不是回到抽象层重构骨架\n- 模型容易为了“做完整”增加无关文件、过多示例或过深目录\n- 模型容易把实现词当成触发词，导致 description 过度技术化\n- 这些是高频失误模式，用来提醒判断边界，不是必须执行的步骤\n\n## 通用判断框架\n\n优先沿着这四个维度判断，而不是直接启动完整流程：\n\n1. 目标和成功标准是否明确\n2. 当前是开放式设计问题、确定性执行问题，还是验证问题\n3. 哪些能力需要 `references\u002F`，哪些动作需要 `scripts\u002F`\n4. 是否真的需要评测、baseline 或触发回归\n\n## 任务判断与分流\n\n开始前先识别当前请求属于哪一类：\n\n### 新建 skill\n\n- 何时归类：用户要把某个工作方式、知识域或工具链沉淀成可复用 skill\n- 核心产物：清晰的定位、顶层结构和 description 草案\n- 优先读取：`references\u002Fauthoring\u002Fpatterns.md`、`references\u002Fauthoring\u002Foutput-patterns.md`、`references\u002Fauthoring\u002Fexample.md`\n- 需要 workflow 细节时再读：`references\u002Fauthoring\u002Fworkflows.md`\n- 需要快速生成骨架时才调用：`scripts\u002Finit_skill.py`\n\n### 融合或改造 skill\n\n- 何时归类：用户要合并两个 skill、重写现有 skill、引入另一份能力，或把旧骨架重新分层\n- 核心产物：改造策略、迁移边界、统一术语和目录组织\n- 优先读取：`references\u002Fmerging\u002Fmerge-playbook.md`\n- 需要写作模式或示例时再读：`references\u002Fauthoring\u002Fpatterns.md`、`references\u002Fauthoring\u002Fexample.md`\n- 只有在确定性迁移步骤已经明确时，才进入脚本或批量调整环节\n\n### 评测与对比\n\n- 何时归类：用户要验证 skill 是否真的提升结果，比较新旧版本，或为改造结果建立 benchmark\n- 核心产物：区分 skill 价值的 eval 集、baseline 选择和可回看的结果\n- 优先读取：`references\u002Fevaluation\u002Fworkflow.md`、`references\u002Fevaluation\u002Fschemas.md`\n- 需要执行或聚合时可调用：`scripts\u002Frun_eval.py`、`scripts\u002Frun_loop.py`、`scripts\u002Faggregate_benchmark.py`、`scripts\u002Fgenerate_report.py`\n- 需要人工判读时可配合：`agents\u002Fgrader.md`、`agents\u002Fcomparator.md`、`agents\u002Fanalyzer.md`、`eval-viewer\u002Fgenerate_review.py`\n\n### 触发优化\n\n- 何时归类：用户要排查为什么没触发、提高触发率、降低误触发，或重写 description\n- 核心产物：清晰的 description、成组的 should-trigger \u002F should-not-trigger 样本，以及回归结论\n- 优先读取：`references\u002Ftriggering\u002Fdescription-optimization.md`、`references\u002Fevaluation\u002Fschemas.md`\n- 需要批量生成或回归时可调用：`scripts\u002Fimprove_description.py`、`scripts\u002Frender_trigger_eval_review.py`、`scripts\u002Frun_eval.py`\n\n不要默认把四类流程全部跑完。只执行当前任务真正需要的部分。\n\n## 资源边界与退化策略\n\n### 渐进披露\n\nskill 默认采用三层结构：\n\n1. frontmatter：只有 `name` 和 `description` 等高频元数据，始终在上下文里\n2. 顶层 `SKILL.md`：只保留哲学、判断框架、边界和资源入口\n3. 按需资源：`references\u002F`、`scripts\u002F`、`assets\u002F`、`evals\u002F`\n\n约定：\n\n- 顶层 `SKILL.md` 实务上尽量控制在 500 行以内\n- 单个 reference 很长时，优先在文件开头补目录或分节导航\n- 只在当前任务需要时才继续读取对应 `references\u002F`\n- 只在步骤已经明确且适合确定性执行时才运行 `scripts\u002F`\n\n## 目录约定\n\n典型结构：\n\n```text\nskill-name\u002F\n├── SKILL.md\n├── references\u002F\n├── scripts\u002F\n├── assets\u002F\n└── evals\u002F\n```\n\n在本 skill 中：\n\n- `references\u002Fauthoring\u002F`：顶层哲学、workflow 取舍、结构模式、示例\n- `references\u002Fmerging\u002F`：融合策略、迁移收尾\n- `references\u002Fevaluation\u002F`：eval、grading、benchmark、viewer\n- `references\u002Ftriggering\u002F`：description 与触发优化\n\n## 工具链前提\n\n- 文档规划、结构设计、融合方案，可以只做文档层规划\n- 自动评测、触发检测、description 优化通常需要本地命令行 runner 或模型命令\n- 脚本默认采用中立接口，不预设特定供应商 CLI\n- 如果环境暂时不支持自动检测，退化为半自动模式：保存输入、输出、日志和人工判定结果\n- workflow 是可选分支，不是顶层默认步骤；只有当顺序执行能明显降低风险时才启用它\n\n## 最小验收清单\n\n- `name`、目录名、skill 定位一致\n- `description` 同时覆盖“做什么 + 何时触发”\n- 顶层 `SKILL.md` 保持精简，且只承担哲学、判断和路由职责\n- 细节已下沉到正确的 `references\u002F` \u002F `scripts\u002F`\n- 路径、命名、术语、归属信息已经统一\n- 高自由度任务没有被误写成固定流程\n- 如果做了融合，结果不是简单拼接\n- 如果做了评测，baseline 合理且结果可解释\n","# Creating Skill\n\n本 skill 用于设计、改造和验证 Agent skill。默认先判断任务性质、产物和边界，再按需读取 `references\u002F` 或调用 `scripts\u002F`，不默认把所有请求推进到完整闭环。\n\n## 定位与适用任务\n\n- 从零设计新 skill，或把一套工作方式沉淀成 skill\n- 重写、分层、融合或重构已有 skill\n- 为 skill 设计评测、baseline、回归检查或结果展示\n- 优化 description、触发边界或排查为什么没触发\n- 如果只是普通文案修改、普通脚本分析或单次交付，不要误用为 skill 创建流程\n\n## 设计哲学\n\n激发模型能力上限 = 策略哲学 + 最小完备工具集 + 必要事实说明\n\n### 策略哲学\n\n- 写思考框架，不写唯一路径\n- 先判断当前任务是开放式设计问题、确定性执行问题还是验证问题\n- 只有当 workflow 能显著降低风险时才引入它；不要把 workflow 当顶层默认范式\n- 高自由度任务必须保留重规划空间，不能被固定流程吞掉\n\n### 最小完备工具集\n\n- 顶层 `SKILL.md` 只放定位、判断框架、边界和资源入口\n- `references\u002F` 提供按需细节，不默认全读\n- `scripts\u002F` 承接重复、确定性、易错的动作\n- `agents\u002F`、`assets\u002F`、`eval-viewer\u002F` 是可选能力层，不是每次都要启用\n\n### 必要事实说明\n\n- 事实说明提供推理原料，例如结构约束、环境前提、触发边界和兼容要求\n- 不要把经验偏好伪装成硬规则；事实说明负责校准判断，不负责替模型决策\n- 当事实边界不完整时，优先补信息，再决定是否进入 workflow 或校验流程\n\n## 工作原则\n\n### 先提取上下文，再补问题\n\n- 如果当前对话里已经包含 workflow、工具链、用户纠偏、输入输出样例或约束，先从现有上下文提取\n- 只有当关键信息仍缺失时，再补问题，不要让用户重复已经说过的话\n- 提问题时优先补这几类空白：触发时机、输出格式、边界、依赖资源、是否需要评测\n\n### token 是共享资源\n\n- 顶层 `SKILL.md` 只保留路由、原则、关键决策和入口说明\n- 长示例、schema、评测细节、大段领域知识放进 `references\u002F`\n- 重复、脆弱或需要确定性的逻辑优先做成 `scripts\u002F`\n\n### 沟通风格随用户熟悉度调整\n\n- 默认优先用用户意图、任务目标和产物来描述，不要上来就堆术语\n- 只有当上下文已经表明用户熟悉时，才直接使用 `eval`、`benchmark`、`assertion` 等词\n- 如果术语不可避免，先用一句中文短解释再继续，例如“评测样本（eval）”或“可验证断言（assertion）”\n- 对熟悉工程流程的用户可以更直接；对非工程背景用户要优先讲目的、步骤和交付物\n\n### 先复用，后新增\n\n- 先盘点现有 skill、脚本、参考资料、模板和历史流程\n- 能改造就不要重写，能下沉就不要堆在顶层\n- 不要为 skill 新增无关文件，如 `README.md`、`CHANGELOG.md`、安装说明\n\n### 自由度和风险匹配\n\n- 高自由度：策略建议、结构设计、边界澄清、示例组织\n- 中自由度：推荐模式明确，但允许按上下文调整\n- 低自由度：容易出错、需要严格顺序、必须稳定复现的流程\n- 高自由度任务禁止被固定流程吞掉；只有风险和收益都明确时才把它压成 workflow\n\n### description 决定是否会被用到\n\n`description` 必须同时说明：\n\n- skill 做什么\n- 什么时候必须使用\n- 用户可能怎样表达这个需求\n\n描述要优先写用户意图和使用时机，再补常见说法、相关文件类型和相邻边界。\n\n## 模型惯性与边界\n\n- 模型容易把开放式任务写成固定路径，即使当前信息不足\n- 模型容易在旧 skill 上持续打补丁，而不是回到抽象层重构骨架\n- 模型容易为了“做完整”增加无关文件、过多示例或过深目录\n- 模型容易把实现词当成触发词，导致 description 过度技术化\n- 这些是高频失误模式，用来提醒判断边界，不是必须执行的步骤\n\n## 通用判断框架\n\n优先沿着这四个维度判断，而不是直接启动完整流程：\n\n1. 目标和成功标准是否明确\n2. 当前是开放式设计问题、确定性执行问题，还是验证问题\n3. 哪些能力需要 `references\u002F`，哪些动作需要 `scripts\u002F`\n4. 是否真的需要评测、baseline 或触发回归\n\n## 任务判断与分流\n\n开始前先识别当前请求属于哪一类：\n\n### 新建 skill\n\n- 何时归类：用户要把某个工作方式、知识域或工具链沉淀成可复用 skill\n- 核心产物：清晰的定位、顶层结构和 description 草案\n- 优先读取：`references\u002Fauthoring\u002Fpatterns.md`、`references\u002Fauthoring\u002Foutput-patterns.md`、`references\u002Fauthoring\u002Fexample.md`\n- 需要 workflow 细节时再读：`references\u002Fauthoring\u002Fworkflows.md`\n- 需要快速生成骨架时才调用：`scripts\u002Finit_skill.py`\n\n### 融合或改造 skill\n\n- 何时归类：用户要合并两个 skill、重写现有 skill、引入另一份能力，或把旧骨架重新分层\n- 核心产物：改造策略、迁移边界、统一术语和目录组织\n- 优先读取：`references\u002Fmerging\u002Fmerge-playbook.md`\n- 需要写作模式或示例时再读：`references\u002Fauthoring\u002Fpatterns.md`、`references\u002Fauthoring\u002Fexample.md`\n- 只有在确定性迁移步骤已经明确时，才进入脚本或批量调整环节\n\n### 评测与对比\n\n- 何时归类：用户要验证 skill 是否真的提升结果，比较新旧版本，或为改造结果建立 benchmark\n- 核心产物：区分 skill 价值的 eval 集、baseline 选择和可回看的结果\n- 优先读取：`references\u002Fevaluation\u002Fworkflow.md`、`references\u002Fevaluation\u002Fschemas.md`\n- 需要执行或聚合时可调用：`scripts\u002Frun_eval.py`、`scripts\u002Frun_loop.py`、`scripts\u002Faggregate_benchmark.py`、`scripts\u002Fgenerate_report.py`\n- 需要人工判读时可配合：`agents\u002Fgrader.md`、`agents\u002Fcomparator.md`、`agents\u002Fanalyzer.md`、`eval-viewer\u002Fgenerate_review.py`\n\n### 触发优化\n\n- 何时归类：用户要排查为什么没触发、提高触发率、降低误触发，或重写 description\n- 核心产物：清晰的 description、成组的 should-trigger \u002F should-not-trigger 样本，以及回归结论\n- 优先读取：`references\u002Ftriggering\u002Fdescription-optimization.md`、`references\u002Fevaluation\u002Fschemas.md`\n- 需要批量生成或回归时可调用：`scripts\u002Fimprove_description.py`、`scripts\u002Frender_trigger_eval_review.py`、`scripts\u002Frun_eval.py`\n\n不要默认把四类流程全部跑完。只执行当前任务真正需要的部分。\n\n## 资源边界与退化策略\n\n### 渐进披露\n\nskill 默认采用三层结构：\n\n1. frontmatter：只有 `name` 和 `description` 等高频元数据，始终在上下文里\n2. 顶层 `SKILL.md`：只保留哲学、判断框架、边界和资源入口\n3. 按需资源：`references\u002F`、`scripts\u002F`、`assets\u002F`、`evals\u002F`\n\n约定：\n\n- 顶层 `SKILL.md` 实务上尽量控制在 500 行以内\n- 单个 reference 很长时，优先在文件开头补目录或分节导航\n- 只在当前任务需要时才继续读取对应 `references\u002F`\n- 只在步骤已经明确且适合确定性执行时才运行 `scripts\u002F`\n\n## 目录约定\n\n典型结构：\n\n```text\nskill-name\u002F\n├── SKILL.md\n├── references\u002F\n├── scripts\u002F\n├── assets\u002F\n└── evals\u002F\n```\n\n在本 skill 中：\n\n- `references\u002Fauthoring\u002F`：顶层哲学、workflow 取舍、结构模式、示例\n- `references\u002Fmerging\u002F`：融合策略、迁移收尾\n- `references\u002Fevaluation\u002F`：eval、grading、benchmark、viewer\n- `references\u002Ftriggering\u002F`：description 与触发优化\n\n## 工具链前提\n\n- 文档规划、结构设计、融合方案，可以只做文档层规划\n- 自动评测、触发检测、description 优化通常需要本地命令行 runner 或模型命令\n- 脚本默认采用中立接口，不预设特定供应商 CLI\n- 如果环境暂时不支持自动检测，退化为半自动模式：保存输入、输出、日志和人工判定结果\n- workflow 是可选分支，不是顶层默认步骤；只有当顺序执行能明显降低风险时才启用它\n\n## 最小验收清单\n\n- `name`、目录名、skill 定位一致\n- `description` 同时覆盖“做什么 + 何时触发”\n- 顶层 `SKILL.md` 保持精简，且只承担哲学、判断和路由职责\n- 细节已下沉到正确的 `references\u002F` \u002F `scripts\u002F`\n- 路径、命名、术语、归属信息已经统一\n- 高自由度任务没有被误写成固定流程\n- 如果做了融合，结果不是简单拼接\n- 如果做了评测，baseline 合理且结果可解释\n",{"name":5,"description":7},"main","d02a86e6a48c1a21cd6e4f492b91fad958d4f128","https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills","https:\u002F\u002Fgithub.com\u002Fdulltackle","experience and enjoy :)","2026-11-06 02:12:13",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add dulltackle\u002Fkangkang-skills --skill creating-skill-pro",{"tool":36,"command":37},"bunx","bunx skills add dulltackle\u002Fkangkang-skills --skill creating-skill-pro",{"tool":39,"command":40},"pnpm","pnpm dlx skills add dulltackle\u002Fkangkang-skills --skill creating-skill-pro","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002FSKILL.md",true,[44,50,56,60,64,70,76,80,85,89,93,97,101,105,109,113,117,120,124,128,132,136,140,144,148,152,156],{"path":45,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":48,"sourceHtmlUrl":49},".gitignore","OTHER","其他",30,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002F.gitignore",{"path":51,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":54,"sourceHtmlUrl":55},"agents\u002Fanalyzer.md","REFERENCE","参考文档",1707,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fagents\u002Fanalyzer.md",{"path":57,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":58,"sourceHtmlUrl":59},"agents\u002Fcomparator.md",1025,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fagents\u002Fcomparator.md",{"path":61,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":62,"sourceHtmlUrl":63},"agents\u002Fgrader.md",1526,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fagents\u002Fgrader.md",{"path":65,"resourceType":66,"resourceTypeDesc":67,"sizeBytes":68,"sourceHtmlUrl":69},"assets\u002Ftrigger_eval_review.html","ASSET","资产文件",13657,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fassets\u002Ftrigger_eval_review.html",{"path":71,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":74,"sourceHtmlUrl":75},"eval-viewer\u002Fgenerate_review.py","SCRIPT","脚本",16200,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Feval-viewer\u002Fgenerate_review.py",{"path":77,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":78,"sourceHtmlUrl":79},"eval-viewer\u002Fviewer.html",45079,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Feval-viewer\u002Fviewer.html",{"path":81,"resourceType":82,"resourceTypeDesc":83,"sizeBytes":84,"sourceHtmlUrl":15},"LICENSE.txt","LICENSE","许可证",11357,{"path":86,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":87,"sourceHtmlUrl":88},"references\u002Fauthoring\u002Fexample.md",923,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Freferences\u002Fauthoring\u002Fexample.md",{"path":90,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":91,"sourceHtmlUrl":92},"references\u002Fauthoring\u002Foutput-patterns.md",953,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Freferences\u002Fauthoring\u002Foutput-patterns.md",{"path":94,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":95,"sourceHtmlUrl":96},"references\u002Fauthoring\u002Fpatterns.md",2482,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Freferences\u002Fauthoring\u002Fpatterns.md",{"path":98,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":99,"sourceHtmlUrl":100},"references\u002Fauthoring\u002Fworkflows.md",2651,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Freferences\u002Fauthoring\u002Fworkflows.md",{"path":102,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":103,"sourceHtmlUrl":104},"references\u002Fevaluation\u002Fschemas.md",3318,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Freferences\u002Fevaluation\u002Fschemas.md",{"path":106,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":107,"sourceHtmlUrl":108},"references\u002Fevaluation\u002Fworkflow.md",2306,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Freferences\u002Fevaluation\u002Fworkflow.md",{"path":110,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":111,"sourceHtmlUrl":112},"references\u002Fmerging\u002Fmerge-playbook.md",1690,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Freferences\u002Fmerging\u002Fmerge-playbook.md",{"path":114,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":115,"sourceHtmlUrl":116},"references\u002Ftriggering\u002Fdescription-optimization.md",4807,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Freferences\u002Ftriggering\u002Fdescription-optimization.md",{"path":118,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":13,"sourceHtmlUrl":119},"scripts\u002F__init__.py","https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002F__init__.py",{"path":121,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":122,"sourceHtmlUrl":123},"scripts\u002Faggregate_benchmark.py",14298,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002Faggregate_benchmark.py",{"path":125,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":126,"sourceHtmlUrl":127},"scripts\u002Fgenerate_report.py",12265,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002Fgenerate_report.py",{"path":129,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":130,"sourceHtmlUrl":131},"scripts\u002Fimprove_description.py",7465,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002Fimprove_description.py",{"path":133,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":134,"sourceHtmlUrl":135},"scripts\u002Finit_skill.py",7395,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002Finit_skill.py",{"path":137,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":138,"sourceHtmlUrl":139},"scripts\u002Fpackage_skill.py",1787,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002Fpackage_skill.py",{"path":141,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":142,"sourceHtmlUrl":143},"scripts\u002Fquick_validate.py",20853,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002Fquick_validate.py",{"path":145,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":146,"sourceHtmlUrl":147},"scripts\u002Frender_trigger_eval_review.py",2359,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002Frender_trigger_eval_review.py",{"path":149,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":150,"sourceHtmlUrl":151},"scripts\u002Frun_eval.py",7267,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002Frun_eval.py",{"path":153,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":154,"sourceHtmlUrl":155},"scripts\u002Frun_loop.py",9478,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002Frun_loop.py",{"path":157,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":158,"sourceHtmlUrl":159},"scripts\u002Futils.py",4585,"https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fcreating-skill-pro\u002Fscripts\u002Futils.py",[161,170,177],{"skillNo":162,"name":163,"nameZh":6,"description":164,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":165,"licenseFileUrl":166,"sourceHtmlUrl":167,"categories":168,"keywords":169,"publishedAt":19},"01M77P7X6TQZ814DJCWBXFDVBK","ppt-master","AI-driven presentation workflow for generating editable PPTX decks and slides, reconstructing page visuals, creating reusable Brand\u002FStyle\u002FLayout\u002FDeck workspaces, filling native PPTX templates, and enhancing finished PPTX files. Use when the user asks to create, generate, reconstruct, regenerate, beautify, redesign, template, fill, or enhance a presentation, PPT, PPTX, slide deck, or courseware — including adding narration or animation to one — requests a presentation-authored narrated\u002Fself-running video, or mentions ppt-master.","MIT","https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fppt-master\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fppt-master\u002FSKILL.md",[],[],{"skillNo":171,"name":172,"nameZh":6,"description":173,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":174,"categories":175,"keywords":176,"publishedAt":19},"01M77P7QPK8QEA8A1XS6N7Z4QK","open-code-review-delegate","Delegation mode for open-code-review (OCR). Instead of OCR calling an LLM endpoint, this skill instructs the host agent to perform the code review itself, using OCR only for deterministic engineering: file selection and rule resolution. Use when the host agent should drive the review with its own LLM capabilities.","https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fopen-code-review-delegate\u002FSKILL.md",[],[],{"skillNo":178,"name":179,"nameZh":6,"description":180,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":165,"licenseFileUrl":181,"sourceHtmlUrl":182,"categories":183,"keywords":184,"publishedAt":19},"01M77P7J3A1QFWFY5JV0N994XP","kami","Typeset professional documents with Kami templates: resumes, one-pagers, white papers, letters, portfolios, slide decks, and social media cards. Use when asked to 做 PDF \u002F 排版 \u002F 简历 \u002F 一页纸 \u002F PPT \u002F slides \u002F 小红书配图, or to create a Kami landing page. Not for auditing or restyling an existing product site.","https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fkami\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fdulltackle\u002Fkangkang-skills\u002Fblob\u002Fmain\u002Fkami\u002FSKILL.md",[],[]]