karanb192/awesome-claude-skills：🎯 这是50+经过验证的Claude Awesome Sk

技术解读

以下内容由 GitRead 中文社区编辑团队基于项目 README 与官方资料整理，结合中文开发者的常见使用场景补充技术选型建议。具体以项目官方文档为准。

深度分析：karanb192/awesome-claude-skills 项目解析

项目背景与痛点

在当今日益复杂的软件开发环境中，开发人员需要一系列高效且经过验证的工具来提高生产力。然而，面对成千上万的开发工具和库选择，找到真正有帮助的工具是一大挑战。项目 karanb192/awesome-claude-skills 正是为了解决这一痛点而生，它为开发者提供了一站式资源集合，涵盖了五十多款经过验证的 Claude Skills（Claude Code, Claude.ai, 和 API 的相关技能）。

这些技能涵盖了从测试驱动开发（TDD）、调试、Git 工作流到文档处理等方面，旨在帮助开发者在多个开发环节中提高效率。该项目作为一个社区驱动且积极维护的项目，确保提供的技能始终保持更新和有效。

核心技术揭秘

项目的核心在于高效管理和组织这些经验证明有效的重要技能。具体实现方式以模块化和可扩展性为设计原则，确保每一个技能的独立和可复用。

graph TD A[用户请求] --> B[目录查询] B --> C[获取特定技能] C --> D[返回技能详情] C --> E{"技能验证"} E -->|"通过"| F[文档展示] E -->|"未通过"| G[返回错误] F --> H[功能解释] F --> I[代码示例]

在数据流方面，用户通过项目提供的HTTP API请求特定的技能。项目后端接收到请求后，首先通过一个目录查询来定位对应的技能，然后对技能进行验证。验证通过后，系统返回详细的技能文档和使用示例。整个流程保证了技能的可靠性和可用性。

功能亮点与差异

相较于市场上的其它竞品，项目 awesome-claude-skills 的杀手锏在于其广泛的技能覆盖面和社区维护的特性。不仅仅是提供技能，它更关注这些技能是否有效且实用。社区驱动的特性使得技能库始终保持与时俱进，确保每个技能都是经过验证且高效的。

且项目中的每一项技能都配有具体的功能解释和代码示例，这不仅方便了技能的学习和应用，还使得开发者更容易上手。

应用场景与落地建议

适用于场景包括任何需要提高开发生产力的团队和项目，尤其是需要处理诸多开发工具选择和实现的复杂业务环境。在生产环境中使用项目时，建议团队根据实际需求对技能进行测试和验证，以确保兼容性和适用性。

综合评价

优点：

广泛的技能覆盖，满足开发者多样化的需求。
社区维护，确保技能库的时效性和有效性。
详细的技能文档和代码示例，便于开发者上手和使用。

缺点：

由于项目依赖社区贡献，更新频率和质量依赖于社区的活跃度。
技能的具体使用场景可能有限，某些特殊需求的开发者可能无法找到满意的解决方案。

总体而言，karanb192/awesome-claude-skills 是一个在技术开发生产力领域不可多得的资源集合，值得每一位开发者在日常工作中参考和应用。

关于

GitHub 原始描述（英文）

🎯 The definitive collection of 50+ verified Awesome Claude Skills for Claude Code, Claude.ai, and API. Boost productivity with TDD, debugging, git workflows, document processing, and more. Community-driven, actively maintained.

agent-skills ai anthropic anthropic-claude artifical-intelligence automation awesome awesome-list claude claude-ai

MIT License
589 Stars
554 Forks
1 Watching

描述取自 GitHub 官方 API，保持仓库作者书写的原文。

简要分析

热度分

1698

价值分

518

活跃状态

活跃

GitHub

贡献者

Find a skill for the task at hand. A community directory for Claude Code, [Claude.ai and the API][anthropic-install].

Browse skills · Install your first skill · Collections · Contribute

Pick a starting point

Your next task	Try this
Extract tables from a PDF	pdf
Find why a test keeps failing	systematic-debugging
Build a feature with tests first	test-driven-development
Create a Word document	docx
Turn a repeated task into a skill	skill-creator

Browse skills

Code	Create	Work
Testing	Files	Workflow
Debugging	Media	Data
Build apps	Writing	Finance & tax
Security	Skills	Automation

Skill names link to their source. Read the install guide there before enabling a skill. Inclusion is not a security audit. Use Cmd+F or Ctrl+F to search.

Document & File Processing

Skill	Use it to
cue-omni-reader	Parse documents and web sources into text through the Cue Omni Reader MCP service.
docx	Create and edit Word documents with tracked changes and comments.
equalang	Translate files or transcribe recordings through Equalang using an API key and credits.
humanpen	Rewrite, translate, shorten, and restyle citations in files through paid HumanPen APIs.
mubu-integration	Import, export, and organize Mubu outlines and Markdown through an unofficial API.
pdf	Extract text and tables, combine PDFs, and fill forms.
pptx	Create, edit, and inspect PowerPoint presentations.
rtl-slides	Build right-to-left Persian and Arabic reveal.js presentations with localized digits, Jalali/Hijri dates, and PDF/PNG export.
translate-book	Translate papers and books with a LaTeX-source route for preserving arXiv equations and tables.
xlsx	Build and edit spreadsheets with formulas and formatting.

Testing & Quality

Skill	Use it to
aeo-check	Lint a site’s code or live URL for what blocks AI search and agents, fix findings in framework order, then re-run to confirm.
agent-qa-authoring	Author and validate agent-qa tests and IDs; source license restricts competing use.
align	Collect structured feedback on generated claims in an HTML form and archive corrections for later review.
benchmark-yylo	Prepare reusable cases, compare models/harnesses and independently evaluate retained outputs.
checkup	Run hosted probes and retrieve server-scored results; full mode uploads approved conversation logs.
clousd	Test Android apps on cloud phones: run plain-language tasks or drive the screen step by step.
crosscheck	Compare independent Claude and Codex analysis, reconcile findings, and verify completed work.
design-fidelity-verify	Compare measured values in a running app with a Figma design specification.
fresh-review	Review code or a plan against acceptance criteria with an agent that has not seen the conversation, then label each finding.
ironloop	Plan Rust work through specifications, tests, simulation, and authorized security checks.
never-again	Capture repaired bugs as enforceable hooks or brief lessons for future sessions.
pdp-audit	Audit ecommerce product pages with desktop/mobile evidence and prioritized UX findings.
playwright-expert	Write and debug Playwright browser tests with fixtures, page objects, visual checks, and CI integration.
prove-it-ran	Show evidence that tests and builds actually ran and covered the change before calling work done.
sim (interview-sim)	Practice timed coding interviews with script-managed clocks and hidden-test grading.
test-checklist	Derive a QA plan from app code with steps and expected results for each platform.
test-driven-development	Write a failing test, implement the change, then refactor.
test-master	Plan and implement unit, integration, performance, and security tests with coverage and defect analysis.
webapp-testing	Exercise a local web app with Playwright and inspect its behavior.
what-could-break	Trace a change through consumers, stored data and duplicated rules, then design a concrete check for the critical assumption.
wreck-it	Test a local web app like real users and chaos monkeys, then report reproduced bugs with screenshots, source locations and regression tests.

Debugging & Troubleshooting

Skill	Use it to
agenttrace-session-audit	Inspect agent sessions for cost, tokens, failures, and latency.
debugging-wizard	Reproduce failures, test hypotheses, identify root causes, and verify fixes.
solidping-monitoring	Find why an uptime check fails, read incidents, and add checks through the SolidPing MCP server.
systematic-debugging	Investigate a bug’s cause before attempting a fix.
verification-before-completion	Run checks and inspect their output before calling work done.

Collaboration & Workflow

Skill	Use it to
a2ui-ask	Collect structured user input through a browser form and JSON answer files.
ai-agency-os-starter	Keep each client’s voice, claims and scope isolated, and check client requests against the SOW before doing unbilled work.
ai-code-review-triage	Review AI-written PRs in three passes: fresh-context bug hunt, context checks (units, history, callers), and a scoped sign-off.
ai-meeting	Run structured proposal reviews with CLI agents and preserve each agent session between rounds.
ax-extract-workflow	Reconstruct a shipped feature’s workflow from local ax session history.
before-you-build	Check demand, distribution, scope, and failure risks before building a product or feature.
brainstorming	Work through requirements and design choices before implementation.
breather	Offer session stopping points with attended-time context and a written handoff.
call-me	Call or text the user’s iPhone to ask or report, then act on the reply; needs the paid /call-me iPhone app.
clueless	Name assumptions, irreversible steps, and review gaps when a user needs extra guidance.
communication-protocol-setup	Agree on assistant communication preferences and export a reusable protocol.
cooper-email	Send, read, and search mail from the agent’s own inbox, and wait for replies or verification codes, through Cooper Email MCP or HTTP.
delegate	Delegate a precisely specified coding task to a worker, then review its diff and tests.
dialog-tree	Track conversation branches in an interactive dialogue tree.
[executing-plans][executing-plans]	Carry out an implementation plan with review checkpoints.
fabling	Apply a work profile for proportionate investigation, visual checks, and task completion.
[finishing-a-development-branch][finishing]	Decide how to integrate finished work and clean up the branch.
forward-deployed-selling	Structure enterprise sales discovery, stakeholder analysis, deal qualification, and account plans.
golden-frijoles	Turn ideas into bets with a target and read date, ship them behind flags and A/B tests, and report whether each paid off.
guashuai / junshi	Coordinate model tiers with either an orchestrator or an occasional expert advisor.
guided-goal	Turn rough intent into a completion contract with success criteria, verification, boundaries and stop conditions before work starts.
karpathy-guidelines	Surface assumptions, keep coding changes small, and define checks before implementation.
kgai knowledge-graph	Capture project decisions and domain knowledge in a linked graph through the kgai CLI.
least-cost-capable	Pick the lowest declared cost rank among models that have the capabilities a task needs.
mindpalace	Organize a local knowledge vault around resources, project pointers, logs, and runbooks.
OrcaReplay	Inspect recorded agent runs and replay or fork them through Orca MCP.
orchestrating-subagents	Choose model tiers for delegated work and review dispatch cost and duration.
planning-with-files	Persist plans, findings, and progress across long tasks and context resets.
pr-review	Filter a pull request review down to actionable findings.
product-decision-agent	Review product priorities, growth problems, and delivery constraints in Chinese.
product-manager-skills	Critique PRDs, diagnose SaaS metrics, plan roadmaps, and run discovery; source license is CC BY-NC-SA 4.0.
punchcard	Review a code change for module boundaries, dependencies, data models, and error paths.
puzzle-activity-planner	Plan puzzle-based activities with timing, difficulty, preparation steps, and links to printable puzzle tools.
[receiving-code-review][receiving-review]	Evaluate review feedback and work through requested changes.
[requesting-code-review][requesting-review]	Request a review before work proceeds or merges.
Satori	Guide reflective conversations about emotions, relationships, and purpose, with professional-care boundaries.
session-handoff	Toolkit for coding agents providing a SKILL.md protocol, context-hygiene checklist, and handoff-note template for structured session transitions.
shipreel	Render a narrated PR walkthrough with diagrams, code panels, and optional app recordings.
simple-man	Shorten agent replies while preserving findings, fixes, checks, and required facts.
slop-post	Create a private Slop draft from a session with a receipt of model, tools, and commits.
tlgr	Read, search, and manage a personal Telegram account through a JSON CLI.
unpaid-intern	Turn meeting transcripts, docs, and tickets into a morning briefing, meeting prep, and follow-ups in plain files you own.
[using-git-worktrees][worktrees]	Work in isolated checkouts for separate development tasks.
using-lwc	Recall and preserve source-grounded project knowledge with LWC wiki and graph tools.
wise-men-flash	Put a hard decision to three subagents, one a Devil’s Advocate, and get one checked memo that keeps the best case against it.
workflow-design	Model workflow states, test guards, analyze transition logs, and plan safe flow changes.
working-memory	Preserve a bounded work checkpoint across context compaction.
[writing-plans][writing-plans]	Break a specification into implementation steps.

Development & Architecture

Skill	Use it to
anti-slop-design	Review and refine UI typography, color tokens, interactions, and accessibility.
anti-ui-slop	Define product-specific UI decisions, cover required states, and review rendered interfaces.
api-designer	Design REST and GraphQL contracts with resource models, OpenAPI specifications, and validation.
aria-icons	Search ~1.4M open-source SVG icons and add them as SVG or React/Vue/Svelte/Solid components
AuraKit	Coordinate build, fix, review, and deployment modes with state snapshots and optional local hooks.
auteur	Build websites from a written art direction and check design consistency and rendered output.
birdview	Map architecture, constraints, and planned changes back to source evidence.
brewpage-publish	Publish text, Markdown, HTML, a file or a folder as a site to brewpage.app and return a shareable link, no account.
build-with-better-design	Select and install a design system through the Better Design MCP service.
building-blog	Add a Next.js and Sanity blog with project discovery, an implementation plan, and SEO guidance.
clone-team	Rebuild website interfaces with an agent team, visual checks, and architecture notes.
cohesivity	Provision and manage application backend services through Cohesivity APIs and MCP.
deacon-install	Add the Deacon AI support widget to a site or app, pass in the signed-in user, and check a live page loads it.
dropthehassle-publish	Publish built static sites through DropTheHassle and connect an existing backend.
easy-auto-research	Run a supervised deep-learning experiment loop with versioned code and persistent progress.
embedded-iot-mentor	Plan embedded and IoT projects, compare hardware, and work from breadboard to firmware.
EmblemAI Agent Wallet	Query wallets and prepare explicitly confirmed transactions through the EmblemAI CLI.
enterprise-architect	Plan target architectures, capability maps, decision records, and technology roadmaps.
erupt-admin	Scaffold a Java admin app with the erupt framework, H2, CRUD, and permission models.
figma-design-extract	Extract Figma values into a design specification mapped to the project tokens.
game-build-team	Build Godot game features with agent roles, headless tests, visual review, and resumable progress.
impresario	Invent a subject-specific concept and build an animated website from one sentence, then screenshot and revise it.
keyboard-shortcuts	Audit and implement web-app shortcuts, command palettes, and keyboard help.
legacy-modernizer	Plan incremental migrations and refactor legacy systems with dependency maps and behavior checks.
light-cloud-deploy	Deploy a repo or local folder to Light Cloud, add a database, and read logs or roll back.
mcp-builder	Build Model Context Protocol servers that connect tools and APIs to Claude.
multi-stack-mcp	Generate hero, pricing, features, and CTA sections for Next.js, Flutter, WordPress, or Vue from shared tokens.
nopekit	Turn physical design claims into checks that must prove they can fail, for CAD, 3D printing, structures, and thermal.
pit-stop	Find a small repository maintenance fix, support it with evidence, and review the patch.
postgres-pro	Analyze PostgreSQL queries, replication, JSONB, extensions, and database maintenance.
regulex-plus	Render JavaScript regular expressions as SVG, PNG, or Mermaid diagrams through a CLI.
scrollytelling-brand-site	Build scroll-told company websites with WebGL print-on-scroll 3D, performance budgets, responsive QA, and an optional CMS.
tastegate	Build a frontend from a design brief, then check layout, contrast and other visible defects with a bundled Playwright browser gate.
tree-ring-memory	Recall, capture, audit, and forget durable project memory through Tree Ring Memory.
unflat	Add visual depth to web marketing pages through a bounded CSS treatment.
web-artifacts-builder	Build web artifacts with React, Tailwind CSS, and shadcn/ui.

Security & Performance

Skill	Use it to
16-eyes	Audit a repository or PR diff through saved investigation lenses and skeptical verification.
awesome-bug-bounty	Plan authorized bug-bounty research with scope checks, vulnerability guides, and report templates.
cra-vulnerability-obligations	Screen CRA scope and vulnerability-reporting questions using cited provisions through Ansvar Gateway.
deep-security-audit	Map code attack surfaces and review candidate vulnerabilities with independent verification.
deep-security-check	Run local static code, dependency, and secret scans and report their evidence and coverage.
gedik	Audit authorized project surfaces and attach reproducible evidence to security findings.
incident-reporting-navigator	Map EU incident-notification questions to cited duties, authorities, and deadlines through Ansvar Gateway.
regulatory-threat-model	Build evidence-cited threat models and selected EU obligation screens through Ansvar Gateway.
SecHelix	Review authorized local code for security issues with evidence and a refutation pass.
secure-code-guardian	Implement authentication, authorization, input validation, and secure coding controls.
security-reviewer	Review code and infrastructure for security issues and produce prioritized remediation guidance.
unthrottled-route-check	Flag Next.js API routes that call OpenAI, Anthropic or Stripe Checkout with no rate limiter before abuse runs up the bill.
wp-security-audit	Audit WordPress configuration, plugin vulnerabilities, and indicators of compromise.

Documentation & Automation

Skill	Use it to
brickwise	Explain a codebase or topic as an interactive 3D brick model in one HTML page.
code-documenter	Write docstrings, API specifications, documentation portals, and developer guides.
devops-engineer	Configure CI/CD, containers, infrastructure templates, deployment checks, and rollback plans.
docs-to-book	Crawl technical documentation and compile an offline PDF book with Typst, optionally translated.
famulor-assistants-history	Read Famulor assistant settings and interaction history through its restricted MCP profile.
linkedin-outreach	Research prospects, draft outreach, and inspect campaigns through Omentir MCP.
llm-wiki	Ingest sources into a linked Markdown wiki, answer with citations, and check consistency across sessions.
okf	Author and validate Open Knowledge Format bundles of linked Markdown concept files.
process-builder	Turn a process interview into a BPMN swimlane diagram in a .drawio file.
project-planning-journaling	Scope a project and maintain a resumable journal of decisions, progress, and open work.
readme-atlas	Write, rewrite, review, and translate project READMEs from repository evidence, dropping claims the repository cannot support.
swarmvault	Maintain a local Markdown knowledge vault with graph navigation, search, review, and optional model-backed synthesis.
trace-my-code	Maintain domain flows, architecture notes, and reuse patterns with code-drift checks.

Media & Content Creation

Skill	Use it to
3d-logo	Build a rotating 3D logo component with React Three Fiber.
algorithmic-art	Create generative art with p5.js.
bria-ai	Generate and edit images or remove backgrounds through the Bria API.
canvas-design	Create visual designs as PNG and PDF files.
cliptude-plan	Plan every shot of a YouTube documentary, then have Cliptude voice, source and render it.
collage-design	Compose collage artwork from licensed or supplied imagery and keep an attribution ledger.
comment-to-dm-cta	Plan the “comment KEYWORD” CTA of an Instagram Reel and create its auto-DM in CommentKeyword, which requires an account and API token.
harmonic-mixing	Look up tempo and musical key to plan playlists and DJ transitions.
hitpost	Write three X posts or threads per idea, length-checked the way X counts, with share links.
identify-song	Identify a track from a link or audio file through Song Finder.
kavel-image	Generate images through Kavel’s anonymous submit-and-poll API within its free allowance; photo edits require a Kavel API key.
keyword-icons	Create several simple SVG icon concepts and inspect transparent PNG renders at small sizes.
ltx2-video	Generate clips from text, images, or video through an LTX-2.3 backend deployed to the user Modal account.
make-faceless-video	Draft, price-check and render faceless short videos through the AutoShort or StoryShort MCP server; needs an account with credits.
meshy-pose-rigging	Prepare character references and work through Meshy auto-rigging and export troubleshooting.
pro-motion	Audio-first motion-graphics launch, brand and TVC films rendered to MP4 in code, cut to a planned music track.
publishport	Publish and cross-post through connected accounts in the user’s PublishPort browser app.
puzzlegenio	Choose a puzzle type and build prefilled links to PuzzleGenio browser tools for creating printable puzzles.
reels-setup	Install and verify a Remotion studio that builds 9:16 story reels cut to a song’s drop.
runapi-cli	Generate media and run model jobs through the RunAPI CLI; requires configured billing and authentication.
ruxi-skill	Build a single-file visual novel from selected book scenes with source-linked choices.
screenbrowser	Produce narrated web-app tutorials through the Screen Browser MCP service.
seedance-25-prompting	Write and troubleshoot Seedance video prompts, reference roles, audio, and shot timing.
slack-gif-creator	Make animated GIFs sized for Slack.
SocialClaw	Schedule and publish social posts through SocialClaw; requires a configured workspace and active plan.
table-sheet	Turn a D&D Beyond character sheet into a play guide and pre-session checklist.
vox-director	Plan and assemble narrated collage videos through paid Atlas Cloud APIs and local ffmpeg.
watch-igaming	Review a video link or local screen recording with transcript and scene-change frames, then build a QA table, UX audit or meeting minutes.
youtube-transcript	Fetch transcripts, video search results, and playlists through the YouTube Transcript API.

Data & Analysis

Skill	Use it to
adhx	Read X posts and articles as structured JSON through the ADHX API.
aligned-delivery	Analyze sprint and delivery metrics, diagnose team workload, and forecast completion dates.
apify-account-mapping	Map the people currently in named roles at each target account through paid Apify actors.
apify-linkedin-post-engagement	Compare LinkedIn post reactions, comments, and shares through paid Apify data.
apify-linkedin-posts-scraper	Collect public LinkedIn post records through a paid Apify actor.
apify-remote-startup-jobs	Fetch structured remote startup job listings through a paid Apify actor.
apify-yc-startup-jobs	Compare published startup salary and equity ranges through a paid Apify actor.
apify-youtube-transcript-api	Fetch YouTube captions and subtitle formats through a paid Apify actor.
apify-youtube-transcripts-llm-training-data	Build transcript datasets with provenance, language policy, error filtering, and paid Apify usage.
apitube-news-api	Search and filter news through APITube, with API authentication, pagination, and error handling.
claude-ecom	Review ecommerce order CSVs for revenue, retention, and margin patterns.
claude-persona	Simulate persona panels and concept interviews, then summarize themes and comparisons for exploratory research.
converly	Configure Converly conversion flows and inspect test events and delivered conversions.
datawebot	Find products across online shops and every seller’s price with direct links through the hosted datawebot MCP server.
formo-analytics	Query Formo product and onchain analytics through MCP, CLI, or REST.
invoice-winning-numbers	Look up Taiwan invoice winning numbers by period from official sources.
market-mirror	Test ads, prices and sales pitches on simulated buyers and get a go / iterate / no-go report.
octav-api	Integrate wallet balances, history and DeFi positions using an API key; selected agent endpoints also support x402 payments.
pandas-pro	Clean, join, aggregate, and transform pandas DataFrames and time-series data.
polymarket-tennis	Build an observe-only tennis market watcher joining market prices to live match scores.
pvr-inox-radar	Map PVR INOX showtimes in India with seats-together counts and travel-time estimates.
shopify-review-triage	Turn supplied Shopify reviews into source-linked incident, friction, pricing, and feature triage.
site-traffic-teardown	Find a site’s competitors and compare traffic, channel mix, keywords, backlinks, and DR, then name three growth opportunities; paid AIsa data.
sql-pro	Write and optimize SQL queries, inspect execution plans, and compare database dialects.
weread-trial-card-list	Scrape WeRead category and rank pages into a deduplicated list of trial-card readable books, exported as HTML and Excel.
youtube-full	Fetch YouTube transcripts, search results, channel data, and playlists through TranscriptAPI; requires an API key.

Finance & Tax

Skill	Use it to
file-itr	Follow a guided walkthrough of India’s income-tax filing portal.
furusato-nozei	Estimate Japan’s hometown-tax donation cap with stated assumptions and official sources.
ipo-offer-document-review	Review Indian IPO offer documents (DRHP/RHP) against SEBI ICDR, LODR, Companies Act and Ind AS for disclosure gaps.
itr-wala	Prepare Indian income-tax returns using document extraction and Python calculations.
open-finance-skills	Review SMB month-end closes, bank reconciliations, 13-week cash forecasts, and 1099 readiness with 20 finance skills.
simmer-tennis-live-gate	Gate simulated tennis market entries on live match state without placing orders.
stipend	Manage a self-custodied USDC wallet and x402 payments with configurable spending policies.
stock-analysis	Analyze listed companies using primary filings, sector comparisons, and accounting-quality checks.
verify-polish-company	Look up Polish company registrations, VAT status, bank accounts, and EU VAT numbers.

Writing & Research

Skill	Use it to
ai-tell-detector	Audit English or Russian drafts for generic writing patterns and suggest precise edits.
berlin-job-search	Search Berlin and German job boards, check German and Blue Card requirements, and draft a Lebenslauf or English CV without inventing claims.
brand-guidelines	Apply Anthropic’s brand colors and typography to artifacts.
business-name-fit	Assess a business name across cultures, languages, and target markets.
buyer-eval-skill	Evaluate B2B software vendors by checking claims, contradictions, unanswered questions, and purchase requirements.
check-blindspot	Check a belief against the evidence: argue both sides, name the fact that would settle it, and flag overconfidence.
diasporic-intelligence	Apply an attribution and consent framework to writing about cultural and lineage knowledge.
fact-check-x-complete	Capture AI answers and citations, compare atomic claims, and preserve evidence during verification.
feynman-pdf	Create illustrated PDF explainers with plain-language analogies and rendered-page checks.
goethecoach	Practise Goethe exam writing, reading, and speaking with rubric-based feedback.
grill	Challenge a decision with an outside model and name tests that could settle the doubts.
grounded	Draft scientific literature reviews with live source discovery and citation checks.
humanize-chinese	Edit Chinese prose with rewriting, phrase cleanup, and local checks; non-commercial license.
humanize-pro	Rewrite or audit prose against a saved voice profile and publishing-channel rules.
humanizer-ru	Edit Russian prose for its audience while preserving facts, voice, and protected text.
humanizer-zh-tw	Remove AI writing patterns from Traditional Chinese text, replace mainland Chinese terms with Taiwan usage, and keep full-width punctuation.
humanizing-writing	Edit prose for generic structure, filler, weak claims, and repeated writing patterns.
internal-comms	Draft updates, newsletters, and other internal communications.
jd-decoder	Read a job posting for real requirements, salary signals, and ghost-listing risk before applying.
kochab	Run a resume-based job search with tailored drafts, interview preparation, and pipeline tracking.
niubiskill	Review business ideas and choose a next monetization test with explicit evidence and assumptions.
no-slop	Review prose for vague attribution, promotional phrasing, repetitive structures, and other writing patterns.
prospeak	Rewrite existing text for professional audiences with English and Italian register controls.
publora-post-ideas	Choose among three social-post angles, then draft the one the user selects.
reading-analysis	Analyse English exam passages, answer choices, and recurring errors. Noncommercial license.
renwei	Edit Chinese manuscripts for syntax, chapter rhythm, factual consistency, and generic prose.
resume-claim-guard	Interviews candidates to lock a verified fact sheet before tailoring resumes to job descriptions.
resume-studio	Maintain a career evidence corpus and tailor resumes to job requirements without inventing claims.
sales-framework	Review sales copy, pitches, and objections against a published persuasion framework.
Say it plainly	Rewrite prose for a reader without changing its claims. Noncommercial license.
seo-translator	Localize content with market-specific registers while preserving keywords, links, and structure.
signs-of-ai	Edit English and Spanish writing patterns and interpret optional analysis-tool results with limits.
structured-gist	Format explanations and process recaps as nested outlines for skimming.
swedish_mentor	Plan Swedish practice with level-matched video, podcast, and speaking exercises.
tldr	Lead with a short summary while preserving full details and critical caveats.
trade-show-prep	Plan a B2B trade show, draft meeting requests, qualify booth leads, and write follow-ups.
unslop	Rewrite prose to remove repeated AI writing patterns while preserving technical content.
upwork-job-hunter	Check new Upwork jobs through the flconsole MCP server, shortlist the ones that fit, and draft a proposal for review.
zh-tw-humanizer	Edit Traditional Chinese prose with Taiwan terminology and protected-fact rules.

Meta Skills

Skill	Use it to
CERA Project Memory	Preserve project session maps and curate a shared index for later conversations.
claude-memory-manager	Organize Claude Code auto-memory with placement rules, update guidance, and index-size checks.
danshari-skill	Audit installed skills and archive redundant ones after user approval, with a restore path.
emulo	Mine local agent-session history into a personal working profile after approving the scope and cost.
feedback	Turn frustration with an agent into an evidence-backed improvement plan that changes nothing until you approve it.
prime-worker	Delegate multi-turn tasks to persistent local prime-agent worker sessions.
reap (skillreaper)	Measure unused loaded context from local agent transcripts and review reversible cleanup.
rule-architect	Generate modular project rules with a shared index and runtime-specific entrypoints.
sijiao-skill	Build a stateful learning tutor with researched lessons, exercises, and spaced review.
skill-creator	Create skills, evaluate them, and refine their descriptions.
skill-freshness	Audit skills across Claude Code, Codex, Cursor and other agents for broken, outdated, duplicate or now built-in ones, with sources.
skill-skill	Write skills that change the output, then lint them against the Agent Skills spec with a health score.
skill-tuner	Refine an installed skill through a small baseline, focused edits, and a rollback decision.
[subagent-driven-development][subagents]	Implement a plan through delegated tasks and review steps.
task-observer	Capture task friction and propose reviewed changes to reusable skills.
template-skill	Start a skill from a minimal `SKILL.md` template.
whetstone	Distil session lessons into reviewable skill proposals with evidence and duplication checks.
[writing-skills][writing-skills]	Write and test skill instructions before distributing them.

Back to categories ↑

Skill Collections

Browse these repositories when you want a related set of skills.

Collection	Focus
Anthropic skills	Document processing, design, development, and skill examples.
Superpowers	Planning, testing, debugging, and development workflows.
adologyai/content-intelligence-plugin	Competitive ad research, brand tracking, and audience analysis through Adology hosted MCP tools.
Affiliate Skills	Affiliate research, content, distribution, and analytics.
Agent Skills English Productivity Pack	Meeting notes, proofreading, research, code review, and commit messages.
agent-pilot-skills	Cross-model review, parallel research, outbound sales, illustrations, and Chinese-platform workflows.
agentsvc.io skills	Speech to text, web pages to Markdown, email checks, token safety, and document text through paid agentsvc.io APIs.
AI sales skills	Company research, sales qualification, and business outreach workflows in Russian.
aigent-OS	Persistent vault memory, session routines, delegation, hooks, and local maintenance workflows.
Baodao skills	Taiwan public-service lookups for invoices, weather, alerts, transit, and public data.
ChatCrystal	Local memory recall and writeback for coding sessions.
Chisanan232/requirement-zero	Review requirement necessity and audit whether existing code still earns its upkeep.
claude-fable-5-skills	Effort calibration, scope control, and subagent orchestration.
Cogny Marketing Skills	Marketing audits and copywriting; live account integrations require a Cogny subscription.
Cowork Tasks	Create and triage a local task board from authorized Cowork connectors; includes prioritization and board-review skills.
CreatorSkills Free Skills	Blog outlines, short-video scripts, show notes, chapters, newsletter planning, and social copy.
dbhq-uk/marketplace	Remote diagnostics, code search, repository checks, research, writing, and work-service integrations.
Design Engineer Skills	Repo setup, component impact, visual QA, reviewable UI pull requests, design review of PRs and design system scoring for designers.
DurdeuVlad/persona-write	Persona-based drafting, rewriting, sample-based voice extraction, and independent review.
E-commerce skills	Review checkout, product content, margin, inventory, retention, and disputes from store data.
Growth Cab GTM Skills	Sending-domain checks, cold-email grading, lead-list QA, and meeting benchmarks.
iOS agents and skills	Swift/iOS implementation, testing, accessibility, performance, security, and release checks.
Kudosity skills	SMS, MMS, WhatsApp, RCS, contact lists, and delivery webhooks through Kudosity.
Kurashi skills	Japan public-service lookups for weather, alerts, holidays, taxes, and libraries.
leechen298/Code2Skill	Extract source-based agent tools and independently review generated workflow and source fidelity.
Lesile-Yin/agent-skills	Prompt engineering, multi-lens reviews, token budgets, and batch PowerPoint automation.
Linked API Skills	Read and manage LinkedIn profiles and lead pipelines through Linked API; requires configured credentials.
Mamba Labs Skills	Prospect research, CRM operations, email deliverability, and go-to-market workflows; some use Apify.
MarchTalk/ai-content-skills	Chinese concept analysis, conversation-to-draft organization, and source-based content review.
Marketing Skills	SEO, copywriting, email, pricing, advertising, and analytics.
mblode/agent-skills	UI, typography, developer experience, documentation, review, and release workflows.
minxnnu-cloud/make-new-things	Claude/Codex visual handoffs, representative frame approval, rework limits, and skill-tree sync.
Nero1688/claude-academic-skills	Academic planning, literature checks, research methods, and replication workflows; mixed licenses.
noizai/skills	Text-to-speech dubbing and companion voice presets.
NotFair	SEO, search visibility, advertising, and analytics workflows with service-specific MCP integrations.
Novu skills	Build Novu notification workflows, inboxes, preferences, and hosted agent channels.
Ontoly skills	Architecture, dependency, refactoring, and documentation workflows grounded in Ontoly graph evidence.
Pathmode	Turn product ideas into intent specifications, challenge assumptions, and verify implementation evidence.
prompt-to-asset	Route image briefs, create software assets, and export platform bundles through a CLI and MCP server.
RouterBase Agent Skills	Integrate the RouterBase gateway, choose models, and handle media-generation API jobs; requires a RouterBase key.
Sequenzy skills	Run lifecycle, campaign, and transactional email work through the Sequenzy CLI or MCP server; requires a Sequenzy account.
Shipwise	Plan and review webapp launches across validation, design, architecture, quality, security, and operations.
sjh9714/skill-receipts	Small code-scope and reproduction-first skills with published admission and rejection evidence.
suede-creator-skills	Code quality, design, marketing, and shipping.
superseo-skills	SEO audits, content writing, and link building.
Vibe Coding Rules skills	Explain errors, add a page, commit, review changes, apply design rules, guard database changes, and run an SEO check.
Wondel.ai Skills	Apply product, UX, marketing, and software frameworks from published books and style guides.
YouTube Skills	Draft YouTube titles, descriptions, hooks, thumbnail briefs, and upload plans; optional Publora publishing.
YYLO skills	Manage tasks, artifacts, benchmarks, and durable work records through the YYLO CLI.

Quick Start

Install Anthropic’s document skills in Claude Code from your terminal.

claude plugin marketplace add anthropics/skills
claude plugin install document-skills@anthropic-agent-skills

Provide a PDF, then ask Claude to extract its form fields. Compare the result with your file. [Read Anthropic’s installation guide][anthropic-install].

Other collections, standalone skills, and updates

Use each source repository’s install guide. Collections may ship as plugins, individual folders, or both. Choose one route for each collection.

For a standalone Claude Code skill, put its folder in ~/.claude/skills/<skill-name>/ for personal use, or .claude/skills/<skill-name>/ for one project. Keep SKILL.md and its supporting files together. See the [Claude Code skill guide][skill-guide].

Follow the [plugin update guide][plugin-updates] for plugin installs. For standalone folders, use the source repository’s update instructions. Cloning this directory does not install the listed skills.

FAQ

What is a skill?

A skill is a folder containing instructions in SKILL.md, with optional scripts and resources. Claude loads it when relevant to a task, or you can invoke it directly. See the [skill guide][skill-guide].

Does every skill work in every Claude app?

The format is shared, but tools and installation differ by host. A Claude Code plugin does not install itself in Claude.ai or the API. Check the source’s supported environments and [Anthropic’s separate install routes][anthropic-install].

Skills, MCP, or project instructions?

You need	Use
A repeatable task or workflow	[Skills][skill-guide]
Access to an external service or tool	[Model Context Protocol (MCP)][mcp-guide]
Conventions Claude reads for a project	[Project instructions][memory-guide]

Are the listed skills tested or audited?

This is a community directory, not a certification program. Read the source instructions and scripts, then try the skill on a task whose result you can check. Report broken links or incorrect descriptions through [issues][suggest].

Can I create or contribute a skill?

Start with skill-creator or the template. To add a listing, follow the contribution guide.

Skill ideas

These are requests for contributions, not skills you can install. Know an implementation? [Suggest its source][suggest].

Browse ideas by category

Category	Ideas awaiting a source
Testing	e2e-testing-skill, snapshot-testing
Debugging	performance-profiling
Development	api-development, database-migration, refactoring-patterns
Security & performance	security-review, dependency-audit, performance-optimization, load-testing
Documentation	documentation-generator, changelog-automation, ci-cd-integration
Media	video-editing-helper
Data	data-visualization, sql-query-builder, csv-processing
Writing	research-assistant, technical-writing

Looking for an entry from an older version of this list?

Some entries now live inside other skills or use a different name.

Earlier entry	Current source
artifacts-builder	web-artifacts-builder
condition-based-waiting	[Supporting guide in systematic-debugging][waiting]
defense-in-depth	[Supporting guide in systematic-debugging][defense]
root-cause-tracing	[Supporting guide in systematic-debugging][tracing]
testing-skills-with-subagents	[Supporting guide in writing-skills][testing-skills]
testing-anti-patterns, sharing-skills	No standalone skill found in the [current Superpowers catalog][superpowers-catalog].

Resources

[Claude Code skill guide][skill-guide]
Agent Skills specification
[How Anthropic designed Skills][skills-deep-dive]
Skills announcement
[Simon Willison on Claude Skills][simon-skills]
awesome-claude-code
awesome-claude
Claude Skills Hub
Claudebin and its source
AgentHub: offers a Chinese-language directory of MCP servers and agent skills.
Hypit video guide: walks through planning and rendering a video from a reference format.
awesome-claude-code-hooks: Related directory of Claude Code hook workflows.
500k.io Skills Bank: Related directory of skills, MCP servers, GPTs, and agent tools.
PolySkill: Skill registry and CLI with public source.
Claude Skills 中文市集: Read Chinese-language tutorials for Claude skills and plugins.

Contributing

Add a skill, fix a link, or improve a description. Read the contribution guide, then open a pull request.

Built with contributions from the [community][contributors], including skills from Anthropic and obra. Maintained by Karan Bansal · [Blog][blog].

License

MIT covers this directory. Each listed skill has its own license.

README 内容过长，已截取前 64 KB。完整内容请前往 GitHub 仓库查看。

暂无评论

awesome-claude-skills

技术解读

深度分析：karanb192/awesome-claude-skills 项目解析

项目背景与痛点

核心技术揭秘

功能亮点与差异

应用场景与落地建议

综合评价

相关仓库

关于

简要分析

贡献者

Pick a starting point

Browse skills

Document & File Processing

Testing & Quality

Debugging & Troubleshooting

Collaboration & Workflow

Development & Architecture

Security & Performance

Documentation & Automation

Media & Content Creation

Data & Analysis

Finance & Tax

Writing & Research

Meta Skills

Skill Collections

Quick Start

FAQ

Skill ideas

Resources

Contributing

License

评论