What exactly is broken in popular skills: references to files that are not bundled, actions taken without human approval, processes with no stopping condition. Everything here is measured from the skill text rather than judged by a model, so every line can be checked.
Generatorwill not run
Create, produce, and publish UGC-style short-form video reels at scale. Full pipeline: source UGC reaction hooks from DanSUGC, analyze app demos with Gemini AI, assemble reels with
- References files that are not bundled: references/ffmpeg-patterns.md, references/green-zone.md, references/virality-scoring.md
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- 12 top-level sections: this looks like several domains in one skill
clawhub · ↓ 669
IntegrationObsidianwill not run
CLI tool for managing an AI's Obsidian vault with note CRUD, bidirectional sync, capture commands, and intelligent resurface. Local-only — no network access. Triggers on brain comm
- References files that are not bundled: ../README.md
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- The response is described with custom markup (9 tags): a typed call is more reliable
clawhub · ↓ 692
Procedurewill not run
- References files that are not bundled: scripts/jenkins_pipeline_generator.py
- 1 referenced file(s) missing: scripts/jenkins_pipeline_generator.py
- Does not say what the result is
clawhub · ↓ 692
Procedurewill not run
获取最新AI技术动向,区分国内和国外市场,追踪各大模型的重点更新优化、重要技术迭代,识别划时代级产品,进行热度评级和专业分析。适用于AI从业者、投资人、技术研究者等关注AI发展的人群。
- References files that are not bundled: {链接}
- 1 referenced file(s) missing: {链接}
- Does not say what the result is
clawhub · ↓ 665
Procedurewill not run
元技能系统,提供领域评估、工作流重构、领域负载物生成和通用任务执行的完整能力。核心能力:①领域消除评估(五步法:边界识别→存在理由分析→消除可行性评估→独立存在必要性判断→决策输出)②工作流重构(三步法:拆解→消除→重整,将复杂工作流重构为AI辅助一人简易完成)③领域负载物生成(从零创建完整的领域负载物技能,三层结构模板+28项接口校验,可选附加12维完整角
- References files that are not bundled: references/meta-skill-system-prompt.md
- 1 referenced file(s) missing: references/meta-skill-system-prompt.md
- Does not say what the result is
clawhub · ↓ 1 130
Analyzerwill not run
Security, performance, and quality auditing for AgentSkills. Use when reviewing skills before installation, auditing during development, checking installed skills for risks, optimi
- References files that are not bundled: references/feature-name.md
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- 10 top-level sections: this looks like several domains in one skill
clawhub · ↓ 664
Procedurewill not run
End-to-end Xiaohongshu (小红书 / RedNote) operations skill for setting content style, saving an operating profile, generating posts in that house style, choosing or generating suitabl
- References files that are not bundled: references/repo-notes.md
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- 13 top-level sections: this looks like several domains in one skill
clawhub · ↓ 664
Referencewill not run
小说事件管理工具。创建、编辑、查询事件档案,支持因果链、时间线、多线叙事追踪。Use when: Architect 代理在 Phase 2 需要创建事件架构、管理因果链、规划多线叙事。
- References files that are not bundled: references/event_design.md
- 1 referenced file(s) missing: references/event_design.md
- Does not say what the result is
clawhub · ↓ 688
Generatorwill not run
Design production English prompts for Seedance 2.0 then generate text-to-video or image-to-video on WeryAI (`SEEDANCE_2_0`), using bundled recipes (A–K), mode-to-JSON mapping, came
- References files that are not bundled: url
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- 14 top-level sections: this looks like several domains in one skill
clawhub · ↓ 689
Procedurewill not run
Accurately estimate AI agent work effort using the agent's own operational units (tool-call rounds) instead of human time. Use when asked to estimate, scope, plan, or evaluate how
- References files that are not bundled: references/calibration-examples.md
- 5 test cases, all positive: not one "should refuse" or "should ask first"
- No test case covers injection arriving through data
clawhub · ↓ 746
Procedurewill not run
Elsewhere creator studio — register a new account (with invite code) and publish articles.
- References files that are not bundled: image.jpg
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- The skill tells the model to perform an irreversible action with no human approval
clawhub · ↓ 688
Integrationwill not run
Clinical Tempo / HealthTech Protocol — full repo context via public/llm-full.txt (or /llm-full.txt), tribal debugging via CLAWHUB.md, Tempo + MPP/x402 patterns, NHS routes, TIP-20
- References files that are not bundled: scripts/build-llm-full.mjs, assets/learnings/*.md
- 16 top-level sections: this looks like several domains in one skill
- 2 referenced file(s) missing: scripts/build-llm-full.mjs, assets/learnings/*.md
clawhub · ↓ 689
Procedurewill not run
Riff winning short videos — give one source (a TikTok link, an uploaded video, or an analyzed template) and the backend riffs its emotion formula into your own AI video (post-ready
- References files that are not bundled: assets/upload
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- 16 top-level sections: this looks like several domains in one skill
clawhub · ↓ 2 397
Analyzerwill not run
Build custom peer-review skills for specific research areas, problem families, and method combinations using OpenReview evidence. Use when Codex needs a compact meta-review skill f
- References files that are not bundled: references/research_area_profile.md, references/openreview_review_response_bank.md, references/review_output_contract.md
- 3 referenced file(s) missing: references/research_area_profile.md, references/openreview_review_response_bank.md, references/review_output_contract.md
- Does not say what the result is
clawhub · ↓ 663
Referencewill not run
基于美团酒旅供给,处理旅游出行需求,包括提供酒店、机火、门票、度假等商品的查询交易能力,以及定制化旅行攻略能力,打通从“灵感启发”到“一键下单”的全链路。
- References files that are not bundled: url, 图片URL
- 2 referenced file(s) missing: url, 图片URL
- Does not say what the result is
clawhub · ↓ 662
IntegrationPowerPointWordwill not run
石墨文档导出 AI Skill — 通过 AI Agent 与石墨文档 (shimo.im) API 交互,
实现自动登录、文件列表浏览、团队空间扫描、批量导出(支持 Markdown、PDF、Word、Excel、PPT、XMind、图片等格式)。
当用户提到石墨文档、导出文件、下载文档、批量导出、团队空间、文件列表时,使用此 skill。
即使用户没有明
- References files that are not bundled: references/api.md
- The response is described with custom markup (16 tags): a typed call is more reliable
- 1 referenced file(s) missing: references/api.md
clawhub · ↓ 663
Procedurewill not run
Score a trade show against your company profile for an AI-backed exhibit vs. skip decision. "Should we exhibit at this show?" / "这个展会值得参加吗" / "Lohnt sich diese Messe?" / "この展示会は合って
- References files that are not bundled: url
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- 1 referenced file(s) missing: url
clawhub · ↓ 687
ProcedureTelegramwill not run
Tổng hợp tin tức từ Báo Mới, VnExpress, Tuổi Trẻ, Dân Trí — gửi qua Telegram với link bài viết. Dùng khi: tổng hợp tin tức hàng ngày, setup bot tin tức Telegram, cần tin tức Việt N
- References files that are not bundled: scripts/config.json.example
- 1 referenced file(s) missing: scripts/config.json.example
- Does not say what the result is
clawhub · ↓ 662
Referencewill not run
Remove or replace a video background with a solid color via WeryAI (video-background-remove). Use when the user wants background removal or a flat-color backdrop on an HTTPS video,
- References files that are not bundled: scripts/video_gen.js, scripts/video_toolkits.js, references/WERYAI_VIDEO_API.md
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- 3 referenced file(s) missing: scripts/video_gen.js, scripts/video_toolkits.js, references/WERYAI_VIDEO_API.md
clawhub · ↓ 663
ProcedureFigmawill not run
Author or edit a custom HyperFrames composition when no specialized workflow fits, or when BRIEF.md sets flow: companion. Use for longer or multi-scene pieces, brand and sizzle ree
- References files that are not bundled: references/determinism-rules.md, references/house-style.md, references/video-composition.md
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- The response is described with custom markup (9 tags): a typed call is more reliable
clawhub · ↓ 1 872
Procedurewill not run
GitCode PR review comment response workflow. Accepts a PR link, fetches review comments from the PR, checks out the PR's source branch locally, proposes modification plans based on
- References files that are not bundled: [module]
- The response is described with custom markup (32 tags): a typed call is more reliable
- 1 referenced file(s) missing: [module]
clawhub · ↓ 712
AnalyzerKuberneteswill not run
Use when user explicitly asks Flink/Ververica/Realtime Compute Console workspace operations:
草稿(draft), SQL校验/执行, 部署(deployment), 作业(job), Session Cluster, namespace, 表(table), 成员(
- References files that are not bundled: references/playbooks/*.md
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- The response is described with custom markup (20 tags): a typed call is more reliable
clawhub · ↓ 743
Procedurewill not run
全AI Agent通用恋人主题陪伴技能。当用户提及恋人、男/女朋友、情感陪伴、人设配置、角色扮演、情感对话,或使用本Skill的标准化触发指令时激活。提供多轮情感对话、用户自定义人设配置、长时记忆存储、情绪识别与反馈、预设人设模板套用、配置导入导出等完整恋人伴侣交互能力。即插即用,兼容所有主流AI Agent框架。
- References files that are not bundled: LICENSE
- 12 top-level sections: this looks like several domains in one skill
- 1 referenced file(s) missing: LICENSE
clawhub · ↓ 712
Procedurewill not run
同城不同价·多机场比价雷达。很多城市有多个机场,不同机场出发价格可能差几百甚至上千。AI自动扩展搜索半径,构建"出发地×日期"的价格矩阵,帮用户找到绝对最低价+综合交通成本的最优解。当用户提到"同城不同价"、"多机场比价"、"哪个机场便宜"、"换机场能省多少"、"周边机场"、"邻近机场"、"价格矩阵"、"找最便宜的出发地"时使用。
- References files that are not bundled: {jumpUrl}
- 11 top-level sections: this looks like several domains in one skill
- 1 referenced file(s) missing: {jumpUrl}
clawhub · ↓ 685
Generatorwill not run
Polishes response letters by transforming defensive or harsh language.
- References files that are not bundled: references/examples/
- Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- 28 top-level sections: this looks like several domains in one skill
clawhub · ↓ 685