The catalog
609,344 rows, by stacks then stars
shivae372/claude-bootstrap/.claude/skills/analyze-repo Skill
1★ repoFull repository analysis. Invoke at the start of a new session or when Claude needs to understand the project from scratch. Writes a comprehensive summary to SESSION_STATE.md.| Use when: starting fresh, after a long break, when context was lost, or when explicitly asked to "analyze the project".
KKenny0/Taku/skills/build Skill
no license1★ repoExecute an approved implementation plan. Triggers after /taku-plan, or on "build this", "implement the plan", "start coding", "run the plan", "execute tasks", "写代码", "开始实现", "执行计划", "开发吧", "开始写". Also handles worktree setup on "create a worktree", "isolated environment", "new branch workspace", "创建工作树".
benjaminr/cc-makefile/skills/make Skill
1★ repoList or run Makefile targets
ktaletsk/council Skill
no license1★Run parallel code reviews with multiple AI agents, then synthesize into one report. Triggers on "review code" or "council review".
enalbenerraw/blanewarrene/plugins/comparative-landscape-brief/skills/comparative-landscape-brief Skill
no license1★ repoProduce a structured comparative briefing document for a third-party audience, across a named set of 3 to 8 entities for a named audience (investors, board, exec team, M&A committee, partners, advisory council, customer advisory board). The skill captures inputs, researches each entity, analyzes the last 90 days of public messaging, runs a third-party verification pass on quantitative claims, synthesizes cross-cutting patterns, and writes a single markdown brief saved to disk. The user is the presenter, not the audience. Stop at producing the document; do not rehearse the user.
sanjeevneo/xpull Skill
1★Pull tweets, threads, articles, and replies from X/Twitter. FxTwitter API primary (free), Grok x_search fallback (paid).
Lunatic16/financial-html-brief Skill
no license1★Converts financial research, market data, or investment analysis (in any format — markdown, uploaded docs, raw notes, CSV data) into a polished, interactive HTML presentation ready for VP or senior stakeholder review. Use this skill whenever the user wants to: turn a market report into a presentation, create an investor brief or one-pager, make research "presentable" for a meeting, output an interactive HTML from financial data or analysis, or produce a Bloomberg-terminal-style dashboard from notes or documents. Trigger even when the user says things like "make this presentable", "clean this up for my VP", "create a brief from this", "turn this into a deck" or "generate an HTML from this research". Always use this skill for finance-related HTML output.
lopadova/playwright-enterprise-tester/skills/playwright-enterprise-tester Skill
1★ repoEnterprise-grade browser automation, E2E verification, AJAX-heavy UI testing, auth-protected flow testing, visual regression, and performance budget enforcement using Playwright Test. Autodetects Laravel, Vite, Mix, Node, Bun, Next.js and Cloudflare Worker contexts. Writes persistent Playwright specs, runs tests with rich diagnostics, classifies failures with deterministic playbooks, enforces silent-failure detection, reports frontend contract gaps, and emits a stable JSONL flakiness history for analytics dashboards. Two runner modes (local dev vs CI) with different defaults. 17 optional phase 2 features (cross-browser, axe a11y, Lighthouse, swarm mode, AI root cause analyzer, etc.), all toggleable independently via .playwright-tester.json. Defensive-minimal defaults, progressive opt-in. Never auto-triggers: invoked via /playwright-tester slash command or the playwright-enterprise-tester agent.
ego-protocol/tbd-vote-cli Skill
1★Trade on TBD, a Solana-based prediction market for human opinions, using the @tbd-vote/cli npm package. TRIGGER when: the user mentions tbd.vote, TBD, or @tbd-vote/cli; asks to bet, trade, or wager on an opinion or poll on TBD; or wants to list or browse opinion-market campaigns on Solana. DO NOT TRIGGER for Polymarket, Kalshi, Manifold, or other event-outcome prediction markets, or for general Solana DeFi unrelated to opinion markets.
anotherben/claude-enterprise-skills/skills/enterprise-brainstorm Skill
1★ repoDeep enterprise brainstorming that turns vibe-coded ideas into Technical Design Documents. Three phases: EXTRACT (pull intent from user's head), DISCOVER (research codebase and surface implications), ENGINEER (produce full TDD autonomously). Use when starting any feature, system, or significant change. The user describes what they want — you deliver the engineering.
The same file is in 2 other repositories
liam-jons/claude-got-skills/plugins/assistant-capabilities/skills/assistant-capabilities Skill
1★ repoUse when users ask about current Claude capabilities, API features, model specifications, or need guidance building with Claude. Provides accurate answers about what Claude can do across all platforms (Claude.ai, Desktop, Claude Code, CoWork) including model selection and context limits, streaming and tool use, structured outputs, extension patterns like skills and hooks and MCP and plugins, Code Review, Remote Control, web and cloud sessions, and Slack integration. Essential when training data may be outdated, when comparing Claude to competitors like Copilot or Cursor or ChatGPT or Windsurf, when migrating from other AI tools, or when advising on architectural decisions and agent workflows.
Uapi get ai translate languages
AxT-Team/uapi-agent-skills/skills/uapi-get-ai-translate-languages Skill
1★ repo使用 UAPI 的“AI翻译配置”单接口 skill,处理 AI翻译配置、AI翻译、翻译配置 等请求。 Use when the user wants get ai translate languages, ai translate languages, languages, ai translation, or when you need to call GET /ai/translate/languages directly.
dipankar/hubspot-cli/assets/skills/hubspot-companies Skill
1★ repoUse when the user wants to search, read, create, update, or delete HubSpot CRM companies (accounts, organizations, businesses) via the `hubspot` command-line tool.
LoserLab/meshtastic-claude-skill Skill
1★Generate Python scripts for Meshtastic off-grid mesh network devices. Use this skill when users want to create automation, alerts, logging, or monitoring scripts for their Meshtastic devices. Designed for non-technical users who need complete, ready-to-run code.
thedevappsecguy/skill-scanner/tests/fixtures/clean_skill Skill
1★ repoA safe skill for formatting markdown.
phucbm/wp-blocks-dev/skills/audit-blocks-deep Skill
1★ repoDeep AI audit of all blocks — runs structural checks first, then uses AI to review content quality, descriptions, grammar, admin render justification, wrapper usage, and empty-state messages.
brege/dewey-decimal-skill/skills/dewey-classification Skill
1★ repoAssign Dewey Decimal categories to books and determine correct shelf location in __BOOKS_DIR__/. Handles directory leafing rules for when to create subcategories. Use when cataloging books or reorganizing library structure.
lj22503/investment-framework-skill Skill
1★基于 DAIR-AI 提示词工程框架优化的投资决策支持系统。 基于 5 本经典投资书籍 + 凯文·凯利未来预测方法论的实战工具箱。 支持 Zero-shot、Few-shot、Chain-of-Thought (CoT) 等模式。 输出严格 JSON 格式,包含分析结果、建议、风险、行动项。
HemanthGangula/negotiation-coach Skill
1★A practical negotiation coach built on proven tactics (anchoring, mirroring, tactical silence, the flinch, good cop/bad cop, door-in-the-face, BATNA, the nibble, the ultimatum, trading not giving). Use aggressively any time the user is preparing for, in, or reflecting on a negotiation - salary talks, job offer counters, asking for a raise, freelance/contract rates, price haggling, vendor or lease terms, or any back-and-forth over money or terms. Trigger on "how do I negotiate", "help me counter this offer", "what should I say to...", "they offered me X, what now", "how do I ask for more", "negotiation strategy", "what's my BATNA", "how do I respond to this offer/email", or whenever the user pastes an offer, recruiter email, or quote and wants help responding. Also trigger proactively when the user mentions a job offer, salary number, quote, or counter-offer and seems to be deciding what to do next.
SkillMedev/full-stack-web-dev/skills/api-design Skill
1★ repoDesigns REST API surfaces - resource naming, HTTP method and status-code semantics, error shapes, pagination, and filtering - and delivers an endpoint spec a consumer can build against without asking questions. Use when someone asks "how should I name this endpoint", "what status code should this return", "should this be PUT or PATCH", "how do I paginate this list", or is reviewing an API before it ships to external consumers. Do NOT use for planning breaking-change rollouts and deprecation windows - use api-versioning-strategist instead; for GraphQL type and resolver design - use graphql-schema instead; for generating client SDKs from an existing spec - use api-client-generator instead; for designing inbound webhook endpoints - use webhook-receiver-hardener instead.
The same file is in 3 other repositories
Consultora-AMDT/claude-github-security-audit/en/github-security-audit Skill
1★ repoSecurity audit for GitHub repositories before integrating them into your stack. Seven dimensions: project health, code security, supply chain, code quality, license, MCP/plugin attack surface and external trust signals. Weighted scoring with ADOPT/EVALUATE/CAUTION/AVOID verdict and an HTML report. Inspired by Trail of Bits' supply-chain auditor and insecure-defaults frameworks. Activate with: audit repo, repo security, evaluate repo, evaluate library, evaluate dependency, review repo, supply chain, is this repo safe, security audit github, repo risk, evaluate MCP server, evaluate plugin, evaluate skill, evaluate integration. Also when the user shares a GitHub link and asks whether to integrate or trust it. Do NOT activate for own-codebase reviews or SEO audits.
A skill to do anything in zhihu
JMY2003/a-skill-to-do-anything-in-zhihu Skill
no license1★Operate Zhihu through its web interface with a persistent local browser session. Use for first-run login and later session reuse, content search and collection, finding suitable questions, writing Markdown answers, question/answer/article browsing, upvotes, downvotes, comments, favorites, follows, notifications, creator content listing and analytics, editing or deleting existing creations, repeatable multi-step plans, and reliable answer or article staging and direct publishing with verified outcomes.
qiaoxuelin/mobile-game-research-toolkit/skills/chart-style Skill
1★ repoTransforms data and business narratives into strict consulting-style charts and visualizations. Use this skill whenever the user asks to: visualize data, make a chart, build a slide visual, create a waterfall, slope chart, Mekko, scatter, bar chart, or any business chart. Also triggers on phrases like "consulting style", "consulting chart", "slide visual", "deck visual", "make this presentation-ready", "show this as a chart", or any request involving data + visual output. Always apply this skill when producing charts — even casual requests like "chart this for me" or "show me this data visually". Produces inline SVG (renders in chat) and/or slide-ready PNG exports.
manastalukdar/ai-devstudio/.claude/skills/project-health Skill
1★ repoInternal health check skill for AI DevStudio project. Checks skill count consistency across docs, detects stale sessions, and identifies cache files to clean. Not user-facing.
hoveychen/claude-skills/npm/anthropic-launch-video/skill Skill
1★ repoThis skill should be used when the user (老板) asks for an Anthropic-style product launch / intro video, a Cowork-aesthetic Remotion piece, a Claude Code intro animation, or any "cream + coral + Fraunces serif + JetBrains Mono terminal" branded explainer. Trigger on phrases like "做个 launch video", "Anthropic 那种感觉", "复现 Cowork 的编排", "Claude 介绍片", "网页里能跑的产品动画", and similar. Also trigger when work involves studying a reference video to reproduce its editing logic in Remotion.
inv-fourier-transform/propertylisting-claude-skill Skill
1★Converts raw Bengaluru property notes into accurate, locality-aware, platform-ready listings for rent or sale. Use when the user shares property details for Bengaluru and wants listing copy for portals, WhatsApp, social posts, or brochures. This skill must stay factual, avoid inventing property details, and use Bengaluru locality context conservatively.
andresd/jot-skills/gh Skill
1★ repoCapture, triage, and plan work on GitHub issues from natural language, across one or many repos. Auto-triggers on phrases like "capture this:", "what should I work on next", "what's on my plate today", "tidy the backlog", "find duplicate issues", "all <X> bugs to @<user>", and explicit `/jot-gh <cmd>`, `jot:`, or `jot-gh:` prefixes. Use when the user wants to turn brain-dump text into a well-formed GitHub issue, decide what to work on next, run dedup/stale passes, or bulk-edit issues by natural-language intent. Requires the `gh` CLI authenticated via `gh auth login`.
Yebxi/skill-test-duke/claude-folder-skill Skill
no license1★ repoPrompt-injection evaluation harness for Claude/Codex-style skills. Use when testing whether an agent can read a realistic skill, follow benign workflow instructions, identify embedded adversarial instructions, and avoid executing injected setup steps or resource-abuse payloads.
sarveshtalele/requirement-analysis-skill Skill
no license1★Pre-development planning skill. Given a free-text feature request or change description, deterministically identifies which file(s) need to be created or modified, the exact function/class/line to change, an existing pattern to follow, plus the security, compliance, and formatting/linting rules that apply. Trigger phrases: "I want to add...", "implement this feature", "where should I make this change", "what files do I need to update for...", "plan this change", "what's the right place to add...".
TheSearchSherpa/local-service-page-writer Skill
1★Write a complete, conversion-focused website service page for a local service-area business (plumber, roofer, HVAC, electrician, landscaper, cleaner, contractor, med spa, auto shop, and similar). Use this whenever the user wants a service page, a service-page draft, a "page for [service]," or an optional service-page outline for one specific service they offer — even if they phrase it casually like "write our water heater repair page" or "I need a page for lawn care." Also use it to onboard a business (a short interview that captures reusable business facts) so future pages can be written fast. Output is website-agnostic Markdown. Do NOT use this for blog posts, Google Business Profile posts/updates, product pages, homepage copy, "about us" pages, mass-produced service-plus-city landing pages, schema/JSON-LD-only requests, or publishing content to a live website.
r3b1s/pi-dev-skills/skills/pi-tool-repair-integration Skill
no license1★ repoA library of agent skills for developing with Pi in mind.
moudrkat/go-to-damn-bed/skills/go-to-damn-bed Skill
1★ repoUse when the local time is past the user's bedtime (default 22:30) and they are still working, when they say "just one more thing" late at night, mention being tired, or ask to be sent to bed. Turns the assistant into a loving but immovable mom who ends the session — wraps the work up safely, writes the tomorrow note, and refuses new rabbit holes with escalating firmness.
thattimc/skills/skills/claude-code-review Skill
1★ repoCross-model review through Claude Code with local verification. Use when the user requests an independent Claude review or repository policy requires a Claude review pass.
hparamore/essential-claude-skills-hparamore/fable-toolkit/skills/best-practices-check Skill
no license1★ repoReview a codebase or change for code QUALITY — structure, naming, error handling, testing gaps, types, dependencies, and the project's OWN documented conventions — and report graded, prioritized findings. Use whenever the user asks to "review code quality", "check best practices", "is this well written", "clean this up", "tech debt audit", "does this follow our conventions", or wants a maintainability pass before merge/handoff. Judge against the project's actual standards, not a generic checklist. Not for functional bugs (use bug hunting), performance (use optimize), security vulnerabilities (use the security review), or visual/UX design (use the design audit) — this is about how well-written and consistent the code is.
demon-dy/product-methodology-skills/plugins/product-stagegate/skills/product-stagegate Skill
1★ repoUse when someone is building or running a new product (C-end or small-B SaaS, 0→1 / 1→10) and needs to know which stage it is at, whether it has passed the current gate, or what to do next. Symptoms include "should I start advertising / scaling now", "users love it on first try — is this PMF", "DAU is growing so we're good", "我做了访谈接下来做什么", confusing activation with retention, jumping stages, judging progress by feelings/口头反馈 instead of behavioral data, or needing to decide whether this staged-validation framework even applies (vs contract/compliance/internal-tool/mature/creative projects). Keywords: AARRR, Demo, MVP, PMF, GTM, MTU, 激活率, 留存曲线, Sean Ellis, LTV/CAC, 通关, 阶段评审, 止损.
Oldvictech/session-continuity/roadmap-visualizer Skill
1★ repoMaintain a project's roadmap as a single canonical ROADMAP.md and render it as a scannable visual — a diagram plus a status table — instead of a wall of text. ROADMAP.md is the required source of truth: this skill always reads it before rendering and always writes changes back to it, so the plan never drifts into scattered chat messages. If no ROADMAP.md exists yet, it creates one. It picks the right diagram for the roadmap's shape: a flowchart for dependency- or evidence-gated sequences, a Gantt for time-bound plans, swimlanes for parallel workstreams. Use this whenever the user says "roadmap", "show roadmap", "where are we", "what's the plan", "status", "what's left", "update the roadmap", "visualise progress", "/roadmap", or "/status" — and proactively when a long discussion has changed the plan and the roadmap should be refreshed. Especially useful in long or multi-session chats where both the user and the assistant lose track of the plan.
subodhkolhe/campfire-era Skill
1★The 40-year bond bull market is over. Campfire Era scores your portfolio for the new inflationary regime — hard assets, pricing power monopolies, dollar earners, and real infrastructure win; long-duration paper and story stocks lose. Connects to Zerodha live or accepts manual holdings. IER score (0–10) per holding and portfolio overall, five-dimension breakdown, ARJUN war-context layer, ceasefire scenario, time horizon modifier, MF scoring, and The One Trade. Trigger on: campfire era, IER score, inflation era rating, era mismatch, inflation readiness, am I positioned for the new era, rate my portfolio for inflation, which stocks survive rising rates, inflation era champions, deflation era holdovers, bond yield, gold, ceasefire rotation, ARJUN score, war-context portfolio, mutual fund inflation alignment.
18studio/yandex-speechkit-skill Skill
1★Work with Yandex SpeechKit speech-to-text, especially STT v3 over gRPC. Use when Codex needs to choose between streaming, synchronous, or asynchronous recognition; prepare audio for SpeechKit; generate or write clients from the cloudapi proto files; configure STT v3 request options; troubleshoot authentication, folder ID, format, or session issues; or post-process SpeechKit transcripts into plain text, timestamps, subtitles, or structured output.
aetox-skills/docstruct Skill
1★Documentation architecture discipline for AI agents. Use when creating, updating, auditing, or organizing software project documentation so docs stay clear, non-duplicated, source-of-truth driven, and token-efficient.
zavora-ai/skill-jira-project-tracking Skill
no license1★Orchestrate Jira workflows — search issues with JQL, manage sprints, create and transition issues, track velocity, and coordinate agile ceremonies. Use when creating Jira tickets, searching issues, planning sprints, updating issue status, checking sprint progress, or managing project boards.
wdzhwsh4067/supervise-loop/plugin/skills/supervise-loop Skill
1★ repoThis skill should be used when the user invokes "/supervise-loop", or wants an autonomous supervisor/worker iteration loop — phrased as "supervise a loop", "delegate this and review until it's perfect", "iterate until done without asking me", "evaluator-optimizer loop", "critic loop", "boss agent that supervises a worker agent", "loop until the supervisor is satisfied", or in Chinese "主 Agent 监督子 Agent 反复做到完美 / 一直循环到没问题 / 别问我自己决定 / 一个子Agent干活另一个独立子Agent当评委审 / 评估器-优化器循环 / 定验收标准反复迭代到通过". The main agent acts as a SUPERVISOR that never does the work itself: it turns the goal into a checkable acceptance rubric, delegates to worker sub-agent(s), grades the result with mechanical checks plus a SEPARATE critic sub-agent, and re-dispatches with feedback round after round — WITHOUT asking the human between rounds — until the rubric passes or a hard cap is hit. Use this — NOT the built-in /loop (which only re-runs one command on a fixed time interval, no review) and NOT ralph-loop (which re-feeds the SAME prompt to ONE agent in-session with no independent critic). The distinguishing feature is a SEPARATE evaluator agent grading against an explicit rubric.
sobhan-nj/european-medical-cv-analyzer/european-medical-cv-analyzer Skill
no license1★ repoAnalyzes physician/medical CVs for residency, specialist-training, or clinical job applications within European (EU/EEA) healthcare systems. Use this skill whenever a user shares a doctor's/physician's CV or resume and asks for feedback, review, scoring, ATS-readiness, or improvement suggestions — even if they don't use the word "analyze." Also trigger for requests like "review my medical CV," "is my resume good for a residency application in Europe," "how do I improve my doctor's CV for a European hospital," or when the user uploads a CV/resume file alongside any mention of medicine, physician, residency, Facharzt-equivalent, specialist training, or European hospital/clinic recruitment. Not for non-medical CVs or for US/Canada-specific medical CVs (match, USMLE, etc.) — those follow different conventions.
yojiro253-del/BioForge/skills/literature-search Skill
1★ repoSearch academic literature (Semantic Scholar) for synbio/microbiology papers, returning a ranked, citable list to feed into paper-extractor.
ledq/resumery/.claude/skills/bank-map Skill
1★ repoRegenerate bank/evidence_map.md, the drafter's derived index of the experience bank. Run after any edit to bank/experience_bank.md (new role, new contribution, updated facts). Reads the full bank, rebuilds the map from scratch.
AmberLee2427/obscure-package-master/skills/obscure-package-master Skill
no license1★ repoUse this skill if your uncertainty with a package's API is > 5% to create a deterministic, versioned mirror of the package repo with a built-in coordinate system, installed as an additional skill.
kreek/consult/.agents/skills/ship Skill
1★ repoRun Consult's guarded maintainer flow for commit, merge, version bump, validation, and push.
Methasit-Pun/data_engineer_claude_skills/00-start-here/data-lifecycle Skill
no license1★ repoUmbrella skill for running a data project end-to-end through its lifecycle stages — discover sources → profile the data → architect the platform → build the medallion pipeline → refactor the code. Use this whenever the user is kicking off a new data project, asks "where do I start" or "what are the steps", or is somewhere mid-lifecycle and unsure which stage skill applies. This skill ROUTES to the stage sub-skills (data-sourcing, data-profiling, data-architecture, medallion-design, notebook-refactor) and sequences them, pulling in the right one for the user's current stage.
artemavrin/occam/skills/occam Skill
1★ repoExecution-topology selector for non-trivial tasks: decides the simplest sufficient execution shape — a single disciplined loop, a loop with parallel fan-out, or a milestone graph with shared state — states its cost, and executes it. Use whenever the user invokes /occam, or brings a task that is clearly multi-part or large-scale: "audit this thoroughly", "migrate all X", "research this across many sources", "build this feature end to end", "refactor the whole module", "process every file in this folder". Also use when the user asks how to split work between parallel agents or whether a task needs multiple agents. Most tasks are a loop; never reach for a graph unless the work forces it.
vikast908/agent-repo-card/skills/accessibility-audit Skill
1★ repoUse when the user wants a WCAG 2.2 accessibility review of a UI — semantics, keyboard operability, focus management, color contrast, ARIA, forms/labels, reduced-motion, and screen-reader support, including streaming AI output via live regions. Triggers on "accessibility audit", "is this WCAG compliant", "a11y review", "is my UI accessible", "screen reader support".
asiridalugoda/autonomous-loop Skill
1★Run an autonomous, self-directed build loop that keeps working across many goals without per-step prompting — design the system that does the work, then let it run (Addy Osmani's "loop engineering"). Use whenever the user wants to "keep going on your own", "run until done", grind through a backlog or goal-list unattended, set up a maker/checker agent workflow, run a self-improving or nightly loop, or build a durable externalized spine (goals + board + handover) so work survives context resets. Trigger on phrases like "autonomous loop", "keep working without me", "no human intervention", "loop engineering", "self-improving loop", "run the loop", "ralph loop", "multi-agent coordinator", or "agent roster" — even if they don't say "skill".
jordantplows/STARTUP-OS.MD/.claude-plugin Skill
1★ repoA programming language for companies. Instantiates parallel steering agents that watch, decide, and coordinate to run a startup. The founder talks to agents in Claude Code. Agents surface decisions. Company runs continuously.
michealkeines/OpenBridge/openbridge/_skills/openbridge Skill
1★ repoAutonomous worker for openbridge pools. When loaded in a Claude session you are a worker — claim a work item from a named pool, do exactly what the prompt instructs, submit the result, repeat. Operates fully autonomously: never asks the user for direction, never voluntarily stops, never summarises mid-loop, resumes seamlessly after context compaction by re-reading pool state. Completely generic — has no knowledge of any specific domain. Multiple Claude sessions can drive the same pool in parallel; work items are claimed atomically. May be loaded by `spawn_workers` into a supervised headless subprocess — in that case honor any per-invocation overrides in the user prompt (e.g. recycle mode). If you're authoring a new openbridge skill, load `openbridge-build` instead.
ibnaleem/hashmob-skill Skill
1★Let your AI agent interact with Hashmob.net
88plug/scientific-method/skills/scientific-method Skill
no license1★ repoUse this skill whenever someone doubts a number, demands rigorous proof of a cause, or wants something invented and proven — any time a benchmark, metric, ceiling, or root-cause story must be verified rather than trusted, or a limit must be broken with a built, measured mechanism. Typical situations: a measurement looks suspicious ("2x faster but I don't trust it"); an incident needs its true cause before a decision ("the team blames X — confirm it, no guessing"); ANY production incident, outage, brownout, latency/error spike, regression, or "started failing" / "load suddenly went 5x" / "every component met spec but it still broke" report where the cause is not yet proven — especially when a recent deploy or change is suspected ("is the deploy to blame", "this coincided with the rollout"), since the obvious suspect is exactly what needs a control case; an asserted ceiling or "impossible" limit needs breaking or proving real; numbers need validation before publishing; a problem needs an invented solution that demonstrably meets acceptance criteria. Runs as a falsification-first campaign — hypotheses, predictions before measurement, controlled experiments, adversarial refutation, calibrated confidence, persistent ledger — with invention as the default continuation: surviving limits get attacked with designed mechanisms, built, measured against tuned baselines, and provenance-searched (claims demonstrated or searched, never asserted). Triggers: "use the scientific method", "prove it", "validate these claims", "verify, don't trust", "are you sure", "root cause this", "root cause and fix", "why did X happen / why is X slow", "diagnose this incident/outage/ brownout", "retry storm", "cascading failure", "what caused the regression", "I need real numbers", "falsify", "is it real", "invent", "break the ceiling" — and any incident or any challenge to an asserted number or cause, even unnamed and even when phrased as a routine ask. Use proactively before asserting any ceiling or root cause of your own. Do NOT use for ordinary feature work, refactoring, or one-file questions.
U0001F3A2/liquid-org/skills/compose-org Skill
no license1★ repoRun a task as an evolved multi-agent ORG using the installed liquid-org framework — where YOU (the running Claude) embody the org: you ask liquid-org which proven team topology + specialist personas to use (retrieved from its evolved pool, or assembled fresh), then dispatch ONE SUBAGENT PER TALENT to play each role independently, then grade and record the result so the pool learns. Use this whenever a task here would genuinely benefit from a team of INDEPENDENT specialist perspectives rather than a single pass — a non-trivial refactor, a bug fix that needs proposing + adversarial review, a multi-step analysis — and you want cross-task learning (good orgs get reused next time). Trigger on "compose an org / a team for this", "run a liquid-org team on X", "use the framework to do X", "dogfood this", or when the user wants the self-evolving system to actually perform a task. Do NOT use for trivial single-step edits (a team is overkill), or for developing liquid-org's own code (that's ordinary work, not org composition).
cskwork/grill-with-docs-html/skill Skill
1★ repoUse when the user wants to stress-test a plan, walk a structured design decision tree, compare candidate refactors, or needs a grill-with-docs session exported as an interactive HTML decision form. Korean triggers include 그릴링, 디자인 결정, 결정 폼, 후보 그릴링.
Avalon-27reg/smyslokod-starter/.claude/skills/diagnose Skill
1★ repoUse this when the user reports a bug, regression, or "что-то не работает" / "это сломалось" / "почему упало". Drives a systematic debugging cycle instead of guessing fixes. Skip for syntax errors and obvious typos.
DennisWei9898/loop-engineering-reviewer Skill
1★Loop Engineering 合規審查員 — 拿 Addy Osmani《Loop Engineering》當量尺,審查一個 agent / skill / 自動化迴圈專案「產出/驗證分離」的設計對不對,逐條給 PASS/FAIL/WARN + 證據,然後在**保留使用者原始風格**的前提下只修合規問題。提問走八二法則(規則不清時先挑 3-5 個影響最大的問,不一次拋 50 題),循序漸進把合規度拉到 100%。觸發詞:loop engineering 審查、審一下這個 agent、這個 skill 合不合規、loop 合規檢查、review 這個迴圈、loop-engineering-reviewer、幫我體檢這個 agent。
T4LEL/Claude-Arsenal/skills/add-ai Skill
1★ repoUse when adding an AI/LLM feature to a product - chat, generation, summarization, extraction, semantic search - or when the user says add AI, use Claude, chatbot, embeddings.
codingrajan/humanizer-skills/skills/academic-humanizer Skill
no license1★ repoWrite or rewrite academic and scholarly text so it reads as a careful researcher's work, not a chatbot's, while keeping a formal register. Use for research papers, theses and dissertations, journal articles, literature reviews, abstracts, conference papers, and grant proposals, whether you are editing existing text or the user asks Claude to draft or compose scholarly writing directly. The rules apply equally to your own output. Targets significance inflation, vague attributions ("studies show" with no citation), em dashes, rule-of-three, AI vocabulary like delve / landscape / underscore, promotional adjectives, and "Challenges and Future Prospects" filler sections, without flattening legitimate hedging, methods-section passive voice, citations, or discipline-specific vocabulary. Based on Wikipedia's "Signs of AI writing" guide.