Discover, test, and download 50+ verified Claude skills and Claude Code skills. Free AI workflow templates, developer tools, and prompt frameworks for Claude, Antigravity, Cursor, and Windsurf.
Official Anthropic skill. Forces a deliberate aesthetic direction (typography, color, motion, layout) before any UI code. Explicitly avoids generic AI UI patterns (overused fonts, purple gradients, cookie-cutter cards).
Generates UI in Nothing Phone’s design language—monochrome, typographic, industrial. Includes dark/light tokens, segmented progress bars, and dot-matrix motifs.
Creates static visual art—posters, designs, PDF/PNG outputs—with proper composition and aesthetic principles (not generic stock-art layouts).
Generative art with p5.js—seeded randomness, flow fields, particle systems. Reproducible outputs from seeds.
Audits designs against 19 professional categories: typography, WCAG contrast, spacing, iconography, navigation, design tokens. Produces a scored report out of 100 with actionable fixes. Supports English and Korean.
Autonomous iteration loop inspired by Karpathy’s autoresearch: modify → verify → keep or discard → repeat. Modes cover code, deployment, content, marketing, sales, research, and design.
Full software development methodology as composable skills: brainstorm → plan → TDD → subagent execution → code review. Bundles 14+ workflow skills.
Packs an entire repo into one AI-friendly file—or generates a structured Agent Skill from any codebase with --skill-generate.
Production-grade engineering skills from the Vue/Vite ecosystem—curated patterns, tooling, and best practices for modern frontend stacks.
Gives the agent real browser automation—navigation, extraction, QA, and research in a live browser context via a local daemon + skill.
Intelligent scraping with automatic strategy selection: curl first → quality gate → stealth browser only if needed → API sniffing → validation. Guides TypeScript Apify Actor productionization.
65+ PM skills across 8 plugins—discovery, strategy, execution, launch, growth. Includes Lean Canvas, JTBD, OKRs, PRDs, pricing, value props, and chained slash commands.
Live Jobs-to-Be-Done interviews (text or voice). Captures switching forces (Push/Pull/Habit/Anxiety), scores opportunities with ODI, exports jtbd.json, messaging angles, and GTM briefs.
Express (15–20 min) or Deep Dive (60–90 min) consulting-style discovery. Frameworks include BCG 10/20/70, Andrew Ng’s playbook, Deloitte AI maturity, and value stream mapping.
GEO-first SEO suite: 25 sub-skills + 18 agents for technical SEO, E-E-A-T, schema, AI search optimization, citability scoring, crawler analysis, and PDF/Excel audit reports.
8-phase pipeline with parallel retrieval, source credibility scoring, critique loop-back, and automated validation scripts. Aggregates Brave, Serper, Exa, Jina, Firecrawl via search-cli.
Full academic pipeline: research → write → review → revise → finalize. 13-agent deep research team, 12-agent paper writing, 7-agent peer review, 10-stage orchestrator. Style calibration and anti-AI-pattern checks.
Replaces sycophantic AI with structured critical analysis. Five modes: FULL, INTERACTIVE (Socratic), TLDR, STEELMAN (debate prep), DECISION (tradeoff table).
Mines your knowledge base (Obsidian vault) for non-obvious connections— inspired by Gwern’s LLM Daydreaming. Multi-agent: Sonnet synthesizer + Haiku critic, recency-weighted pair sampling.
Converts 15+ source types (YouTube, PDFs, web pages, articles, WeChat, search queries) into NotebookLM outputs: podcasts, presentations, mind maps, quizzes.
Vector-powered semantic file search across your codebase or documents. Ask natural-language questions like “where did I save brand colors” and get ranked matches.
Multi-pass editing to remove AI-writing tells: em-dash overuse, throat-clearing openers, rule of three, synonym cycling, promotional inflation, and 24+ documented patterns (Wikipedia AI-writing guide).
Hard-edged writing style contract for timeless, forceful English prose—concrete, verb-forward, no therapeutic tone or marketing cadence.
43+ marketing skills in 7 pods: Content (8), SEO (5), CRO (6), Channels (6), Growth (4), Intelligence (4), Sales (2)—plus orchestration router and Python tools.
CRO, copywriting, SEO, analytics, growth engineering, ad creative, content strategy—40 skills with short-form video frameworks for Reels/TikTok/Shorts.
55K+ word email marketing knowledge base (908 sources)—lifecycle flows, segmentation, deliverability, win-back sequences, 19 industry playbooks.
Pulls competitors’ ads from ad libraries, analyzes messaging and creative patterns for teardown before campaign launch.
Avatar extraction, offer extraction, Schwartz awareness mapping, mechanism builder, headline matrix, objection crusher, ad angle multiplier, full-funnel orchestrator.
25 skills grounded in published frameworks—Don Norman (UX), Cialdini (persuasion), Eric Ries (lean), Alex Hormozi (offers), plus product strategy and growth.
17-skill content system: voice, posts, reels, thumbnails, carousels, scoring, analytics, niche research, infographics—newsletter-first, multi-channel output.
Interviews you, analyzes writing samples, produces about-me.md and voice.md—the foundation file for all other content skills.
Reverse-engineers outlier Reels via Apify + Gemini Flash, then writes a new script in your voice from newsletter/source material.
Pulls your post history via Apify and scores drafts against what actually performed for your account—catches voice/performance drift pre-publish.
Turns a video title into a branded YouTube thumbnail prompt for Gemini/image models—CTR-oriented composition patterns.
Generates hooks across PAS, AIDA, BAB, STAR, and SLAY frameworks—the 3-second scroll-stop problem.
40+ X/Twitter actions: post, extract, monitor, compose, schedule, follower export, DMs, giveaway draws—via Xquik REST API and MCP. TweetClaw is the OpenClaw/plugin runtime; x-twitter-scraper is the standalone skill package.
Publishes full Markdown articles to X/Twitter Articles via browser automation—formatting, images, tables (as PNG), Mermaid diagrams, threading awkward bits handled.
Analyzes and rewrites tweets using Twitter’s open-source recommendation insights—Real-graph, SimClusters, TwHIN, Tweepcred—for reach mechanics, not vibes.
Analyzes live public opinion on Reddit and X—real quotes, sentiment, trends. Structured JSON + dashboards, not hallucinated summaries.
Official Remotion skill—programmatic video in React. 30+ rule files: 3D, animations, TikTok captions, FFmpeg, audio visualization, Lottie, light leaks.
End-to-end AI-native video workspace—voiceover scripts, demo recording, Remotion Studio review, ElevenLabs voiceover, MP4 render.
Full-lifecycle AI music—lyrics, Suno-style prompts, per-stem mixing, mastering, release distribution guidance.
Multi-modal generative media for AI agents—image, video, and audio across 100+ models (Kling, Veo, Flux, Suno, etc.). Schema-driven scripts with JSON output for agent workflows. Cross-platform (Claude Code, Cursor, Gemini CLI).
OpenAI GPT Image 2 with thinking modes (off/low/medium/high), 14 style presets, 8 platform presets (Instagram, Story, YouTube, Pinterest), editing and variants.
~286K words of color science—OKLCH/OKLAB, palette generation, WCAG/APCA contrast, color naming, pigment mixing, color theory history.
Generates Excalidraw-style hand-drawn diagrams from prompts—architecture, workflows, UX blueprints. Hosted edit link, animated SVG, PNG export.
27 plugins covering Three.js, GSAP, React Three Fiber, Framer Motion, Babylon.js, A-Frame, PlayCanvas, PixiJS, Locomotive, React Spring, Magic UI, Lottie, Blender, Spline, Rive.
Analyzes medical reports, tracks health metrics, gives personalized wellness suggestions from your actual data—not generic advice.
Personal genome analysis from raw 23andMe or AncestryDNA data—17 categories: health, ancestry, nutrition, pharmacogenomics, longevity. HTML dashboard output.
Build browser-native presentation decks with a mandatory Theme Gallery picker UI, keyboard navigation, reveals, presenter notes, Recharts data/board themes, and exportable slide formats.
Build production-ready prompts for any use case using structured prompt engineering frameworks. Covers writing, coding, marketing, image generation (Flux, Midjourney, GPT-4o), and vibe coding tools (Cursor, Lovable, Bolt, v0).
Everything developers, creators, and teams need to know about discovering, installing, and building with verified Claude skills. Supercharge your agentic workflows across Claude Code, Google Antigravity, Cursor, and Windsurf.
In the rapidly evolving frontier of AI-assisted engineering, Claude skills represent a fundamental leap beyond basic chat prompting. Rather than re-typing verbose system instructions into every terminal session, Claude agent skills package domain expertise, precise behavioral constraints, deterministic tool allowances, and execution workflows into standardized, reusable modules defined by the open SKILL.md specification.
Specifically tailored for Anthropic's official terminal assistant, Claude Code skills empower autonomous agents to execute complex, multi-turn coding routines—from full-stack refactoring to architectural audits. When paired with structured AI workflow templates, these skills allow Claude to analyze codebases contextually, preserve state across tool executions, and deliver production-ready code with minimal human intervention.
PromptSkillForge was engineered as the leading open-source Claude skills marketplace and verified Claude skills directory. Finding high-quality, safe, and effective skills used to require scouring disparate GitHub repositories and unverified gists. Our directory solves this discovery bottleneck by curating, auditing, and indexing 50+ battle-tested workflows into a searchable catalog.
Every entry in our collection provides authentic, verified configurations. Whether you need to download Claude skills directly into your browser with zero terminal dependencies or copy plug-and-play CLI commands, our platform guarantees that you receive the best free Claude Code skills without license friction, hidden paywalls, or unvetted scripts.
Modern AI agents are only as capable as the directives that guide them. Our curated repository categorizes free Claude skills across five mission-critical domains to address real-world production demands:
Our specialized Claude skills for developers and Claude skills for web development streamline modern full-stack workflows. Whether executing pixel-perfect responsive CSS using our Frontend Design skill, scaffolding reactive applications, or configuring Vite bundles, these templates enforce modern software architecture and type safety out of the box.
Ensure bulletproof reliability using dedicated validation modules. The Claude testing skill systematically drafts comprehensive unit and integration suites (Jest, Vitest, Playwright), while the Claude code review skill identifies security vulnerabilities, edge-case race conditions, and performance bottlenecks prior to merging pull requests.
Unlock higher search rankings using Claude skills for SEO automation. From automated technical audits and structured JSON-LD schema generation to localized programmatic keyword mapping via the GeoSEO engine, these skills transform Claude into an organic search strategist.
Supercharge conversion rates with the battle-tested Claude copywriting skill. Created for writers and founders, our Claude skills for content creation handle long-form newsletter development, direct-response sales copy, viral hook generation, and voice matching without sterile AI clichés.
Transform raw datasets into actionable executive insights with Claude skills for data analysis. In tandem with the Claude workflow automation skill, agents can bundle sprawling codebases using Repomix, execute recursive web research, and synthesize statistical summaries automatically.
If you are new to autonomous AI assistants, adopting Claude skills for beginners is straightforward and rewarding. Traditional prompt engineering often suffers from drift—where Claude forgets instructions as context windows fill up. In contrast, understanding how to use Claude skills unlocks deterministic, persistent agent behaviors.
To use any skill from our directory, simply navigate to the skill detail page, copy the accompanying Starter Prompt Template, and paste it into your active AI session. Claude immediately reads the underlying rule set, activates the necessary sub-skills, and proceeds with laser focus on your requested objective.
We believe installing agent workflows should take seconds. Our platform offers multiple methods explaining how to install Claude skills based on your development environment:
~/.claude/skills/<skill-name>/SKILL.md for persistent access across all projects..agents/skills/), Cursor (.cursor/skills/), and Windsurf workspaces.For in-depth walkthroughs and troubleshooting, explore our step-by-step Universal Agent Setup Guide.
Once you have mastered using pre-built templates, discovering how to create Claude skills allows you to codify your proprietary team practices. Creating Claude Code custom skills follows a clean, declarative markdown standard compliant with official claude code skills documentation:
Specify the skill name, concise description, trigger keywords, and operational scope at the top of your markdown file.
Provide clear guidelines: when the agent should trigger this skill, what tools it is permitted to call, and strict rules to avoid common pitfalls.
Include standardized input templates and expected output schemas so you and your team can invoke the custom skill consistently.
Clear, expert answers to questions about finding, testing, installing, and building custom Claude Code skills.
Claude skills are standardized, modular instruction packages (typically packaged as a SKILL.md markdown file with YAML frontmatter) that enhance Anthropic's Claude Code and compatible agentic IDEs. Rather than repeatedly typing lengthy prompt instructions into every conversation, a Claude skill gives the agent persistent domain expertise, predefined operational boundaries, workflow scripts, and deterministic tool rules (such as terminal execution, file system modifications, and browser interactions). All Claude skills on PromptSkillForge are open-source and 100% free to download.
Claude skills transform Claude from a generalized conversational LLM into a specialized, task-focused autonomous agent. When activated during a development session, a skill:
Based on developer adoption, utility, and ecosystem downloads across PromptSkillForge and the Anthropic community, the top 10 skills for Claude are:
The skills you should give Claude depend directly on your role and active project goals:
Pro Tip: Avoid overloading Claude with too many overlapping skills at once. Giving Claude 2 to 4 complementary skills per workspace yields the highest reasoning accuracy and prevents rule conflicts.
You can find verified Claude skills across three primary platforms:
#claude-skills, claude-code, or install via the CLI registry using npx skills add <repo-url>.Trying Claude skills takes less than 60 seconds using any of these simple methods:
SKILL.md content or provided Starter Prompt Template, and paste it into a new Claude chat window or Claude Desktop session. Claude immediately adopts the directives and constraints.npx skills add anthropics/skills --skill frontend-design -g -y to install globally, or paste the skill into your project's ~/.claude/skills/ directory.SKILL.md directly and place it in your workspace at .agents/skills/<skill-name>/SKILL.md (for Antigravity/Windsurf) or .cursor/skills/<skill-name>/SKILL.md (for Cursor), then invoke the skill normally.For detailed step-by-step terminal instructions, explore our complete Universal Agent Installation Guide.
Yes! Anthropic provides an official meta-skill named Skill Creator (skill-creator) specifically designed to instruct Claude on generating new, production-ready Claude skills. It guides Claude on how to formulate valid YAML frontmatter (with title, description, and tags), define unambiguous operational rules, structure trigger criteria, set parameter schemas, and package executable prompt templates.
You can also ask Claude directly: "Create a custom Claude Code SKILL.md for [your specific workflow] adhering strictly to the official Anthropic skills specification."
Creating custom skills for Claude Code follows a clean, declarative markdown standard compliant with official Anthropic documentation:
.claude/skills/my-custom-skill/ and add a SKILL.md file.---
name: my-custom-skill
description: Concise summary of what the skill accomplishes and when to trigger it
tags: [frontend, testing, typescript]
---To quickly learn the fundamentals of Claude, Claude 3.7 Sonnet, and Claude Code:
/help, /init, /compact, and /cost.