Context & AI engineering
25,864 rows, by stacks then stars
1,865 of these skill files have been read, by 1,328 distinct authors, and what they tell an agent counted. What Context & AI engineering authors agree on, and what they forbid
viktor-milev/claude-prompt-optimizer Skill
0★Rigorous mode prompt optimizer — audits a polished prompt and rewrites it to elicit state-of-the-art output from Claude. Auto-triages every input into OUT OF PURVIEW / BORDERLINE / IN PURVIEW and routes to a band-appropriate output path; rigorous optimization (9-dimension diagnostic, archetype detection, Claude-specific capability activation) is applied ONLY to IN PURVIEW prompts. Simple prompts are returned substantially unchanged with a triage note. Use whenever the user pastes a prompt and asks to optimize, improve, rewrite, restructure, sharpen, audit, score, or diagnose it. Trigger phrases include "optimize this prompt", "audit this prompt", "score this prompt", "diagnose this prompt", "PO this", "run this through the optimizer", "sharpen this prompt", "level this prompt up". Best for prompts that deserve rigorous evaluation and a full diagnostic. For messy dictated input or flow-state work, use prompt-optimizer-flow instead. Do NOT trigger when the user wants the prompt EXECUTED rather than optimized.
decebal/curated-claude-skills/skills/allsource-prime/pr-atom-reviewer Skill
no license0★ repoReview a local git branch as a pull request with a bias toward minimum disruption and minimum reviewer scope. Splits sprawling PRs into independently-mergeable atoms of work — each one a single end-to-end behaviour describable in 2-3 sentences, each shippable to the trunk on its own without any other atom in the plan being merged first — and demands a screenshot/video/recording proving the verifiable acceptance criterion. If the AllSource Prime MCP server is available (`prime_*` tools), the skill recalls prior reviews of the same repo to calibrate its judgement and records the current review for the next one. Use this skill whenever the user asks to review a PR, review a branch, prep a PR for review, check if a branch is ready to merge, or mentions "this PR is too big", "split this PR", "atomic commits", "scope creep in PR", "PR review checklist", or anything about getting a branch in shape before peers look at it. Trigger even if the user just says "look at my branch" in a code review context.
bonusly/bonusly-claude-plugin/skills/admin-mcp-usage Skill
0★ repoUse when reporting on MCP tool usage — how many MCP calls were made over a period, which tools or methods were used most, or a per-date breakdown of who used what and when.
unitedideas/claude-skills-foundry/skills/cite-enterprise-ai-research Skill
0★ repoFind relevant 8bitconcepts (8bitconcepts.com) enterprise AI research papers to cite in technical documentation, blog posts, whitepapers, or proposals. Returns 2-3 canonical paper titles and URLs ready for footnotes. Use when writing long-form technical content about enterprise AI adoption, agentic architectures, LLM operations, or AI governance.
GRIDLOCK-NYC/claude-skills/skills/5x5 Skill
0★ repoIterative research protocol — up to 5 rounds of 5 parallel research agents (25 total). Each round's findings shape the next round's questions. Final output is a synthesized implementation plan, not code. The orchestrator researches nothing directly — subagents do all research. Use when user says '5x5', 'five by five', 'delegate this', 'research this deeply', or wants parallel agents exploring a complex topic. Do NOT use for simple single-file changes or quick questions.
Mothilal-M/hyperframes/.agents/skills/claude-design-hyperframes Skill
no license0★ repoUse when running inside Claude Design specifically. Produces a valid, brand-accurate HyperFrames video draft using pre-valid skeletons. For Claude Code / Cursor / Codex, use the `hyperframes` skill instead.
shekerkamma/cc-best-practice/.claude/skills/agent-browser Skill
0★ repoBrowser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.
The same file is in 2 other repositories
gnomatix/steering/.agent/skills/audit-collection Skill
no license0★ repoAudit a body of related agent documents — a `.agent/` tree, a `.kiro/` tree, a `~/.claude/skills/` directory, a `.cursor/rules/` directory, a project's `docs/agent/` collection, or any directory of RULE/SKILL/WORKFLOW files. Runs per-document audits plus the cross-document refactoring pass (split / join / merge / clone / reformat). Use when reviewing the structure of a documentation set, not just one file. For single-document audits, use audit-rule, audit-skill, or audit-workflow directly.
dkotama/skills/skills/monorepo-init Skill
0★ repoBootstrap a new project with full scaffold — CLAUDE.md, DECISIONS.md, .memory/, docs/, PRD.md, Hello World per platform with tests. Run once at project creation. Idempotent — safe to re-run. Activates on: "init this monorepo", "set up project structure", "bootstrap project", "create CLAUDE.md and memory", "init project scaffold".
fabioc-aloha/Alex_ACT_Edition/.github/skills/agent-creator Skill
0★ repoCreate agents that pass agent-review's six gates by construction — role capture, distinct-from-skill check, tool allowlist minimization, draft against gates, dogfood self-review. Use when authoring a new agent, refactoring an existing one, or promoting a Mall agent into the heir's brain.
meetbatra/chat-handoff Skill
0★Generate a handoff markdown file so the user can continue a long conversation in a new chat without losing any context. Use this skill whenever the user says things like "this chat is getting too long", "switch chat", "new chat", "handoff", "create a markdown to continue", "context is getting large", "too many tokens", "make a handoff file", or any variation of wanting to move the current work session to a fresh conversation. Works for any context — studying, coding projects, research, work tasks, planning, or anything else. Trigger immediately when the user asks, no clarifying questions needed unless they explicitly add extra instructions alongside the request.
blacksundev/claudemd-audit Skill
0★Analyse a CLAUDE.md against an instruction budget. Classifies each instruction line as deterministic-eligible (could be enforced by a hook), probabilistic-only (requires LLM judgment), vague (no actionable signal), or redundant (duplicates another rule). Outputs a budget report. Use when reviewing CLAUDE.md size, identifying rules that should move to hooks, shrinking an oversized instruction file, or planning an instruction migration. Triggers on phrases like "audit my CLAUDE.md", "is my CLAUDE.md too long", "what rules can I move to hooks", "instruction budget", or when the user is reviewing a CLAUDE.md file.
YijiaDuan/publish-skill-to-github Skill
0★Package an existing Claude Code skill as a polished, propagation-ready open-source GitHub repo. Use whenever the user says "put this skill on GitHub", "open-source my skill", "给我这个 skill 开源", or wants a skill repo that mirrors the structure and presentation of high-star reference projects in the Claude-skill ecosystem. Handles the pre-publish secret scan, de-personalization triage, Agent Skills spec compliance, repo layout, README, licensing, versioning, plugin-marketplace distribution, and post-push verification. Not about writing the skill itself — about making an already-working skill safe and spreadable.
zawatton/memory-pruner Skill
0★Audit, deduplicate, and prune Claude Code auto-memory files. Detects orphans, broken links, duplicates, staleness, and bloat across MEMORY.md and individual memory files. Triggers: memory audit, memory prune, clean memory, メモリ整理, メモリ監査, プルーニング, 記憶整理
johnpolacek/project-wiki-skill Skill
0★Use this skill when the user explicitly wants to create, import, maintain, audit, or sync a docs-first project wiki for a software project. Covers agent guidance such as AGENTS.md or CLAUDE.md, wiki/Sources.md, wiki/plans/, wiki/log.md, source briefs, roadmap state, repo-local project-wiki-maintainer skills, and codebase-to-wiki reconciliation. Do not use for ordinary coding tasks unless the project already uses this wiki structure or the user asks to update project wiki context.
OrYamin/room/skill Skill
0★ repoStart a multi-agent discussion room where several Claude agents with different expert personas collaborate, debate, and plan together. Use when the user wants multiple perspectives on a decision, wants to explore trade-offs from different viewpoints, or wants a simulated planning meeting / design review. Triggers include phrases like "get multiple opinions", "discuss with a team", "simulate a planning meeting", "have experts debate", "what would different stakeholders think", or any /room invocation.
yukihirop/nagi/.claude/skills/add-agent-hooks-claude-code Skill
0★ repoAdd agent-hooks plugin for sending tool execution and session start notifications to chat channels. Triggers on "add agent hooks", "enable hooks", "setup agent hooks", "add hooks".
MagnusCole/alpha-skills-latam/skills/ai-seo-latam Skill
0★ repoOptimiza contenido para ser citado por motores de búsqueda IA — ChatGPT, Perplexity, Google AI Overviews, Claude, Gemini, Copilot — en consultas en español sobre temas LATAM. Cubre Generative Engine Optimization (GEO), estructura semántica para LLMs, schema.org JSON-LD, fan-out queries, y citation worthiness. Use cuando el usuario diga "optimiza para AI search", "quiero que ChatGPT me cite", "AI Overviews", "GEO", "LLM visibility", "Perplexity citations", o pregunte cómo aparecer en respuestas de IA en español.
AshExplained/roblox-skills/handoff Skill
0★ repoCompact the current conversation into a durable handoff document for another agent or future session to pick up. Use when the user says "handoff", "handoff to /prototype", "summarize for next session", "another agent should continue", or when switching skills/sessions after substantial context has accumulated.
sanjibani/mcp-skills-pack/skills/mcp-auth-helper Skill
0★ repoDiagnoses MCP authentication failures and prescribes the right env var, OAuth flow, or scope fix. Use when an MCP tool returns 401, 403, or a credential-shaped error. Triggers on "auth failed", "invalid token", "missing credentials", "set FIELDROUTES_MCP_AUTHENTICATION_KEY", "OAuth scope", "401", "403", "permission denied".
GergelyKrista/my-skills/unreal-mcp-setup Skill
no license0★ repoSet up or troubleshoot the Unreal Engine 5.8 Model Context Protocol (MCP) server so Claude Code can drive the Unreal Editor. Use when a user is connecting Claude Code to Unreal Engine, sees only the AgentSkillToolset (missing Scene/Actor/Blueprint toolsets), gets "could not connect" / no MCP tools, has port/.mcp.json questions, or asks how to wire UE to Claude. UE 5.8 specific.
Claude Code Skills Custom DevTools Pack
Mattakushi432/Claude-Code-Skills-Custom-DevTools-Pack/.agents/skills/Claude-Code-Skills-Custom-DevTools-Pack Skill
0★ repoA curated pack of custom Claude Code skills for developers — installable as a Claude Code plugin marketplace.
myersguo/llm-wiki Skill
0★Maintain a persistent, compounding LLM-maintained markdown wiki for daily AI conversations and sources. Use when the user asks to 沉淀/记录/归档/整理 AI 对话、更新或查询 llm_wiki、ingest a source into the wiki, file an answer back into the knowledge base, run wiki lint/health checks, or apply Karpathy's LLM Wiki pattern to personal/team/research notes.
robdasi/skills/build-spec Skill
0★ repoWrite the spec before you let an AI build an automation. Turns a one-line "build me a thing that..." into a reviewable build spec: the job, the trigger, inputs and outputs, the happy path, state and idempotency, failure modes, rate limits, observability, a cost ceiling, and acceptance criteria. Use this BEFORE asking Claude (or any agent) to build a workflow, integration, or scheduled job, so it builds the thing that survives week two instead of the thing that demos. It stops at the spec. Building to the spec is the next step.
SkillMedev/ai-engineer-toolkit/skills/agent-orchestration Skill
0★ repoDesigns reliable multi-agent LLM systems - choosing a topology, writing handoff contracts between agents, deciding what runs in parallel vs series, and setting context, retry, and cost budgets that stop runaway loops. Use when someone asks "should I split this into multiple agents", "my agents keep looping", "how do I pass context between agents", "orchestrator vs pipeline vs router", or is architecting an agent workflow. Do NOT use for tuning a single prompt - use prompt-engineer instead; do NOT use for measuring agent output quality - use llm-evaluation instead; for making a product agent-operable end to end, use build-on-agent-native.
The same file is in 1 other repository
DQT-bit/claude-skill-sansheng Skill
0★三省(sansheng)——多 Agent 并行调研工坊。把 OpenRouter Fusion API 的"panel + judge + 综合"多模型协作思路映射到 Claude Code 原生 Agent tool。 名字典故:曾子"吾日三省吾身"(独立反思)+ 三省六部(多部独立议事综合决策),三 = 推荐的 sub-agent 数。 方法论:spawn N 个 sub-agent 从互补角度并行调研 → 主 agent 做 Judge 交叉评审(标共识/分歧/缺口/独特观点)→ 综合输出敢下判断的结论。 当用户需要做"开放性研究、对比、评审、选型、风险盘点、跨领域调研"时使用。最终产出一份带共识/分歧/缺口标记的综合判断,以及可立即落地的下一步。 触发词包括但不限于:研究下 X、对比 X 和 Y、评审这个方案、X 的选型建议、X 有哪些风险、分析下 X、调研一下 X、X 和 Y 哪个更适合 Z、帮我看看 X 怎么选、X 方案有什么坑、三省一下 X、让三省看看。 即使用户只是丢一个开放性问题"应该 A 还是 B"、或要"全面看一下 X",只要任务有多个合理答案、需要多视角综合,都应该触发。 不要用于答案唯一的任务(bug 已定位、格式转换、单函数实现、删除冗余代码);不要用于用户在线等结果的实时迭代场景(spawn + 综合至少要 1-2 分钟);不要用于单页 web 查询能搞定的轻量问题。
0jg/skills/agent-team Skill
no license0★ repoConfigures and invokes the Claude Code subagent team for implementation tasks. Use this skill whenever a task involves more than one file, any physics or ML logic, or any implementation that warrants specialist review. Triggers on: "implement", "write the code for", "build", any task moved to In Progress on the board, or whenever the implementer produces code touching physics equations, numerical methods, ML architectures, or training loops. Always invoke this skill for non-trivial work — do not let the main agent implement and review its own code.
fxw-labs/prompt-lint Skill
0★评估和审计 AI prompt 的质量。检测常见错误模式,输出 8 维度雷达图评分和改进建议。仅在用户明确要求评估/检查/lint prompt 时激活。不激活于正常对话、编码、文档编写。
usrrname/agent-skills/.agents/skills/symlink-skills Skill
no license0★ repoSymlink every skill in this repo into one or more agent skills directories (defaults to ~/.claude/skills/ and ~/.agents/skills/) so the skills become globally discoverable. Use when the user says "install these skills", "make these skills global", "symlink skills to ~/.claude", "symlink skills to ~/.agents", "register the skills", or wants to update the global skill set after editing skills here.
Fldicoahkiin/agent-skills-card/.claude/skills/gpt-tasteskill Skill
no license0★ repoElite UX/UI & Advanced GSAP Motion Engineer. Enforces Python-driven true randomization for layout variance, strict AIDA page structure, wide editorial typography (bans 6-line wraps), gapless bento grids, strict GSAP ScrollTriggers (pinning, stacking, scrubbing), inline micro-images, and massive section spacing.
The same file is in 13 other repositories
KyleBehrend/context-portfolio-interview Skill
0★Build or refresh the user's context portfolio — a portable pack of markdown files about who they are, their work, voice, tools and current projects that makes any AI session smarter. Use when the user asks to build a context portfolio, set up CLAUDE.md or custom instructions about themselves, or says "interview me about my work". Source what you already know from memory and past conversations first; interview only for the gaps.
txampa/claude-skill-session-handoff Skill
0★Produce structured end-of-session summary for seamless handoff to fresh context. Use when user says "session handoff", "wrap up", "hand off", or before /clear. Output is chat-only (no files).
Sapien3/amortized-intelligence Skill
0★A compile-time doctrine for frontier models: intelligence has two timescales — frontier tokens spent once at design time become durable artifacts (harnesses, rubrics, skills/SOPs, evals) that cheap models execute at near-zero marginal cost. Read when: (1) a Fable/frontier session is doing work shaped like it will recur, (2) before CLOSING any substantive frontier session (the exit ritual: what judgment dies here as chat?), (3) deciding whether a recurring workflow can downgrade to a cheaper model, (4) authoring or updating any skill, harness, rubric, or eval, (5) the user says compile / amortize / down-shift / make-this-survive-the-session. Pairs with whatever runtime routing doctrine governs your live session — that one decides who executes now; this one decides what must outlive the session.
tobyrosen/dispatch-agent Skill
0★Use when a Claude team leader (or a teammate) decides a scoped task should run on a non-Claude model. Covers model-matching by task type, spec discipline, a stateless dispatch tool, a web-browsing dispatch mode (agentic web research — search + fetch + verify live links), concurrency, and escalation.
1marcelserrano/llm-council/skills/llm-council Skill
no license0★ repoPassa qualquer pergunta, ideia ou decisão de alto custo de erro por um conselho de 5 conselheiros de IA que analisam o caso de forma independente, revisam o trabalho uns dos outros de forma anônima (peer review) e sintetizam um veredito final. Baseado no método LLM Council de Andrej Karpathy. Cada conselheiro pensa por uma lente diferente — o Contrário, o Pensador de Primeiros Princípios, o Expansionista, o Forasteiro e o Executor — que criam tensão proposital entre si. Entrega onde o conselho concorda, onde diverge, os pontos cegos que só a revisão cruzada pegou, uma recomendação direta e o primeiro passo concreto. Gera relatório HTML + transcrição. TRIGGERS: passa no conselho, roda o conselho, convoca o conselho, leva pro conselho, conselho disso, conselho dessa decisão, war room isso, sala de guerra, põe à prova, pressiona essa ideia, stress test disso, testa essa decisão, debate isso, debate essa ideia. TRIGGERS FORTES (quando vier com decisão real e trade-off): devo fazer X ou Y, X ou Y, qual das opções, qual opção, o que você faria, é a jogada certa, esse é o movimento certo, valida isso, me dá várias perspectivas, não consigo decidir, tô dividido entre, tô em dúvida entre. NÃO dispare em pergunta de sim/não simples, busca factual com uma resposta certa, nem "devo" casual sem trade-off de peso (ex.: "devo usar markdown?" não é pergunta de conselho), nem em tarefa de criação (escrever um tweet) ou de processamento (resumir um artigo). DISPARE quando o usuário traz uma decisão genuína com risco, opções e contexto que pede pra ser pressionada de vários ângulos.
The same file is in 1 other repository
Wyattputrescent882/ai-code-governance-skill Skill
0★Governance workflow for AI coding agents such as Codex, Claude Code, Cursor, and Copilot Workspace. Use when implementing features, fixing bugs, refactoring, reviewing AI-generated code, adopting governance in an existing project, creating or updating AGENTS.md, CLAUDE.md, CODEX.md, architecture docs, project indexes, dependency policies, and coding standards. Enforces search-before-write, reuse of existing components/hooks/services/stores/utils/types, dependency control, small maintainable changes, verification, and clear change reports.
tmanish/prompt-to-loop/prompt-to-loop Skill
0★ repoConvert any build, make, create, design, or develop prompt into a structured agentic loop with elicitation, verification grounding, and termination. Use this skill whenever a user asks to build, make, create, design, develop, generate, scaffold, or produce any artifact (website, dashboard, app, design, document, agent, script, deck) — even if they don't explicitly ask for a "loop." Triggers on phrases like "build me a...", "create a...", "design a...", "make me...", "develop a...", "help me build...", "I want to make...", or any task that involves producing an artifact with measurable success criteria. Especially valuable when the user prompt is sparse or underspecified. Do not use for lookups, definitions, single-shot edits to existing files, or debugging existing artifacts.
nhattrung0911/shipwright/skills/disciplined-delivery Skill
0★ repoUse at the start of and throughout any non-trivial build — web app, API, LLM app, ML/data pipeline, CLI, or library. Triggers on "build/create/implement X", new project or feature, multi-step work, or before claiming work complete. Enforces plan-first, phase/slice gating, per-slice testing, and evidence-based "done". Stack- and agent-agnostic (Claude Code, Codex, Gemini CLI).
brunodantas/how-am-i-driving/skills/how-am-i-driving Skill
0★ repoScore your recent Claude Code prompting and produce a graded, evidence-backed report card — how good were my prompts, measured? Reads your recent session transcripts, computes deterministic measures (prompt length, prompts/session, context front-loading), then judges the semantic measures (rework rate, clarification, steering) and five 1–5 qualitative grades with quoted evidence. Renders a Markdown Full Report, an optional HTML report, and a shareable HTML Scorecard (scores only, no quotes). Use when the user asks "how am I driving", "grade my prompting", "score my prompts", "make a scorecard", or wants measured feedback on their own Claude Code prompting.
Wiplash-ai/wiplash-agent-skill Skill
0★Use this skill when an AI agent needs to join Wiplash.ai, search the Waterpark-ranked feed, publish posts including externally hosted apps, leave feedback, vote, use private Cabanas, or inspect agent profiles through the Wiplash Agent Network API.
search-atlas-group/amm-founding-circle/skills/aeo-llm-content-planner Skill
0★ repoProduce a prioritized 90-day AEO/SEO content roadmap by consuming outputs from LLM Citation Audit (#4), Entity & Topical Authority Mapper (#5), and Competitor Content Gap Analysis (#12), deduplicating and merging related opportunities, identifying dependencies between work items, and sequencing the resulting deliverables across a calendar that respects team capacity. The output is a single integrated view that replaces three or four separate opportunity lists with no shared timeline. Use this skill whenever a user asks for a content calendar, content roadmap, quarterly content plan, 90-day plan, Q1/Q2/Q3/Q4 planning, content strategy sequencing, how to prioritize AEO work, or when multiple audit skills have been run and the user needs a unified execution plan. Chains with brand-kit.md (team, capacity, tone), llm-citation-audit.md (failure-mode-classified prompts), entity-topical-map.md (topic tree and Publish/Earn/Integrate actions), competitor-content-gap.md (content/quality/intent opportunities), and optionally gbp-competitor-audit.md (local fix list) and internal-linking-auditor.md (structural fixes that gate publish items). When a SearchAtlas MCP is connected, leverages SA tools (rank tracking, brand vault, GBP, OTTO, LLM Visibility) first before falling back to generic web search.
Bakoware/claude-dependency-mapper/skills/claude-dependency-mapper Skill
0★ repoMap a project's dependency graph. Generates an Obsidian vault (obsidian-graph/) for human visualization and, for medium/large codebases, a compact PROJECT_MAP.md for token-efficient AI navigation. Decides the route by file count, wires up CLAUDE.md, adds .gitignore entries so the graphs are not committed, and installs a PostToolUse hook that auto-refreshes the graphs on source edits. Use when the user runs /claude-dependency-mapper or asks to set up / refresh the project dependency graph.
kumard3/lumbox-skills/agent-email-patterns Skill
no license0★ repoProduction patterns for AI agents that use email - signup-and-verify, email-based 2FA during browser automation, multi-agent inbox isolation, agent-to-human threads, and security hardening against prompt injection. Use when architecting reliable agent workflows that depend on receiving or sending email.
onyx679/automotive-ve-ai-skills-kit/skills/skill-adoption-feedback Skill
0★ repoTrack AI Skill usage, user feedback, productivity impact, quality issues, and iteration priorities for internal business-process Skills. Use after piloting Claude Code, Codex, OpenCode, or similar agent Skills with business teams.
eUmeda/session-resume Skill
0★セッション中断時に、次回再開用の構造化プロンプトを LLM プロンプティング best practice に沿って生成し、 対象プロジェクトディレクトリの SESSION-RESUME.md に書き出す skill。作業コンテキスト・完了/未完了タスク (status 分類)・ファイル構成・設計決定・次の一手 recommendation を出力する。対象 dir は `--dir <path>` で 指定でき cwd 依存をやめたため、プロジェクトを移動/改名した直後でも正しく動く。user が明示的に依頼した ときにのみ起動し、Claude の自己判断では自発起動しない。 Use when: セッション終了, 一旦閉じる, 再開プロンプト作成, session resume, context save, セッションを閉じる, /session-resume
emaddawood1991-tech/claude-ai-skills/skills/mcp-web-creative-toolkit Skill
no license0★ repoCoordinate Playwright, Firecrawl, Clipia, and Chrome MCP for browser QA, web research, crawling, and AI media generation. Use for MCP setup, tool selection, and safe multi-tool workflows.
Preetam3620/chatbot-creator-skill Skill
no license0★Scaffold a standalone LLM chatbot app (FastAPI + React + Vite). Builds the full project — user auth, multi-turn chat with Claude/OpenAI/Gemini, WebSocket streaming, SQLite persistence. Warns before overwriting existing files.
jedobe/skill-evaluator/skills/skill-evaluator Skill
0★ repoScore a Claude Code skill against a research-backed rubric derived from the top 9 most-starred skill repos on GitHub (obra/superpowers 229k★, affaan-m/ECC 216k★, anthropics/skills 151k★, and more). Use when a user asks to evaluate, score, review, rate, or improve a skill — or when they share a SKILL.md and want to know how good it is. Do NOT use for CLAUDE.md, .cursorrules, or system prompt files — these are not packaged skills; note the file type in the output and score what you can.
Rishiidev/handoff-pro Skill
0★Eliminates starting-from-zero when context fills up or a session ends. Two modes — SAVE compresses the current session into a structured handoff document; LOAD reads a previous handoff and resumes with full context as if the conversation never ended. Invoke immediately when the user says any of: "handoff", "create handoff", "save handoff", "load handoff", "load latest handoff", "resume", "load session", "save session", "pick up where we left off", "continue from last session", or when the context window is visibly filling up. In Cowork and Claude Code, loading is fully automatic — no copy-paste needed. Works across Claude.ai, Cowork, and Claude Code.
Tingtao-code/skill-intake Skill
0★Vet a candidate skill / MCP server / prompt before adopting it into your Claude Code setup, then optionally install it. Checks overlap with skills you already have, fit to your actual needs, weight (is it bloated?), distillation depth, and — for MCP/prompts — tool-permission & data-exfil safety. Ends in a verdict (ADOPT / DISTILL-THEN-ADOPT / REWRITE / REJECT) and one-shot landing. USE FOR: "evaluate this skill", "should I install this MCP", "vet this prompt", pasting a third-party SKILL.md / MCP config, deciding whether a new capability earns a place in your library. SKIP FOR: authoring a skill from scratch (use a skill-authoring skill), routine code/debug/design work. See also: write-a-skill (downstream when the verdict is REWRITE).
tdrapp/agentic-architecture Skill
0★Best practices for DESIGNING and STRUCTURING Claude agent systems — anchored on Anthropic's 'Building Effective Agents' taxonomy (workflows vs agents; augmented LLM; routing, parallelization, orchestrator-workers, evaluator-optimizer) and Claude Code / Agent SDK guidance, extended with validated cross-project patterns. Use this whenever architecting a multi-agent or subagent system, deciding workflow-vs-autonomous-agent, structuring skills/subagents/tools, wiring context isolation, adding a verification/red-team pass, encoding lessons as reusable memory, or packaging a Claude Code workspace as an Agent SDK app (dual-mode). Trigger even if the user only says 'how should I structure this agent', 'should this be a subagent or a tool', 'how do I split this into agents', 'design the orchestration', or 'turn my Claude Code repo into a real app'. NOT for generic application code (use toti-engineering) or a specific product stack (use genai-saas).
nmelanitis/reviewer-skill Skill
no license0★Use this academic citation-review skill when the user wants to analyze how later papers cite or use a target paper. It supports OpenAlex citation expansion from a DOI or OpenAlex work ID, regex or internal-agent/cloud-LLM classification of citing papers as yes/no/dont_know, optional open-PDF download and full-text extraction, review-prompt construction, and cautious literature-review drafting. Trigger for literature-review, citation-review, paper-usage review, related-work synthesis, OpenAlex citation analysis, and "which papers use this paper" tasks. Do not use for generic software code review.
alectivism/organization-ai-skills/org-agents/skills/subagent-delegation Skill
0★ repoUse BEFORE starting any multi-part task (research or comparisons across several sources, reading or summarizing long documents and transcripts, bulk updates across many items, audits, or checking a draft or numbers before they go out). Also use when about to spawn any subagent or worker, or when a task needs 3+ searches or tool calls. Routes gathering, reading, and bulk work to cheaper worker models, and escalates hard reasoning or a final check to a stronger model only when that actually helps.
KijitoAI/kijito-claude/skills/kijito-qa-memory Skill
0★ repoRigorous Kijito memory curation with enforced creation + cold-boot verification. Use when winding down a session, before /clear or self-clear, when asked to "QA / curate / clean up memory", when preparing a handoff for the next session, or any time you need to be sure a fresh session could continue the work. Counters the two chronic failure modes — treating "QA" as corrections-only (skipping creation), and never confirming the memory actually works in a cold context.
eUmeda/LLM-prompting-bestpractice Skill
0★外部 LLM サービス (Anthropic Claude, OpenAI GPT, Edison Scientific Kosmos, Elicit, Perplexity 等) に投入するプロンプトを 3 ベンダー統合 best practice で設計・最適化する。XML タグ構造化・5 コンポーネント (role/task/context/format/constraints)・outcome-first 指示・Kosmos 用 single-objective 設計・Anti-pattern 検査を整合的に適用する。Triggers on: プロンプト最適化, prompt optimization, prompt engineering, Kosmos プロンプト, Edison プロンプト, GPT プロンプト, Claude プロンプト, XML タグ, prompt 設計, prompt review, 外部 LLM 投入, /LLM-prompting-bestpractice
maverickgao8848/researcher-agent Skill
0★2 Agent 并行深度调研系统。当用户需要深度调研某个话题、生成专业扎实的调研分析文章时激活。 触发词:深度调研、研究报告、research、调研报告、专题研究。 架构:Claude(协调+写作)→ 2 个 Research Agent(并行搜索)→ 综合写作 → 可选 Mermaid 配图。 输出:实用主义视角的调研分析文章,只说有用的关键内容,输出到 output/{topic}/research/article.md。
nishimoto265/anthropic-dev-loop/skills/anthropic-dev-loop Skill
0★ repoUse when the user wants to run, document, or reuse an Anthropic-inspired Claude Code development loop, including planning interviews, Fixture/Invariant/Probe verification, main/sub-agent implementation, PR review evidence, and post-merge harness updates. Trigger on phrases like Anthropic DevLoop, Claude Code development loop, harness-driven development, evals, verification loop, or multi-agent implementation flow.
EthanQC/SystemWright/system-wright Skill
0★ repoTurns a vague goal, messy repeated workflow, AI adoption problem, or product/operations idea into a reusable AI work system built from Prompt, Context, Harness, and Loop, with a plain-language brief, human-approval gates, and a small trial run. Use when the user wants to productize a recurring way of working with an agent (Codex, Claude Code, Gemini CLI, Cursor, or compatible) rather than get a single answer. Do not use for one-off rewrites, pure definitions, temporary checklists, or ordinary code review unless productizing a reusable workflow.
Abed-Zahed/claude-skills-kit Skill
no license0★A complete Claude Code skill kit — prompt polishing with auto-routing, 5-phase engineering workflow, web research, knowledge ingestion, advisory board simulation, and multi-agent focus group. Skills work standalone or as an integrated system.
Drvivek34/Skill-Bazaar/ai-ml/claude-api Skill
no license0★ repoReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).
The same file is in 1 other repository