Discover the Most Powerful
AI Skills

Create, modify and A/B test custom Claude Skills with built-in performance tracking and an evaluation framework — your skills keep getting smarter.

GitHub's official MCP server: manage repos entirely from Claude — commit code, open PRs, handle issues and branches without leaving the conversation.

Connect directly to your Supabase project: operate PostgreSQL, Auth users, Storage and Edge Functions in plain natural language.

Privacy-first real-time web search with AI summaries and citations. Regular and deep-research modes, no tracking.

High-quality web search + page crawling + GitHub code context retrieval — built for technical research and code-related queries.

Operate PostgreSQL in natural language: automatic schema detection, complex queries, analysis and result visualization.

Read Slack channels, send messages and search history from inside Claude — handle team comms without switching apps.

Full local file system read/write: create, edit, organize and batch-rename files — Claude works directly on your machine.

Read, search and draft Gmail, manage Google Calendar, Docs and Sheets — one-stop Google Workspace integration inside Claude.

Turn any content into professionally formatted .docx files: automatic tables of contents, heading levels, headers and footers — fully Word-compatible.

Create, edit and analyze .xlsx files: formulas, pivot tables, charts — and clean up messy raw data.

Access Linear issues, projects and docs right inside Cursor — get task context without window switching and ship decisions faster.

Deep Cursor–Figma integration: design-to-code, component recognition and generation, pixel-perfect implementation.

A payments integration assistant with built-in Stripe API knowledge: generates payment flow code and knows webhooks, subscriptions and one-time payments inside out.

Deploy and manage AWS services from Cursor — Lambda, S3, EC2 and RDS in one place, no console switching.

Bring user behavior analytics into the editor — let real usage data guide feature development and optimization decisions.

Manage Cloudflare CDN, Workers, KV storage and security rules from Cursor — the edge developer's best companion.

One-click deploys to Vercel with preview environment management and instant rollback — the whole pipeline without leaving the editor.

Explore MongoDB collections, optimize queries and analyze data in Cursor — natural language to MongoDB queries.

Bring production monitoring into your dev flow: error logs, performance metrics and alerts next to your code for fast incident triage.

GitLab repos and CI/CD in one flow: commit code, watch pipeline status and manage merge requests from Cursor.

ChatGPT's built-in Python environment: upload CSV/Excel and analyze, plot and compute statistics — nothing to install locally.

ChatGPT-native image generation: text to high-quality images with style control, inpainting and multi-image comparison.

Create Canva designs from a ChatGPT conversation: describe what you want and get professional posters, decks and social graphics.

An automation engine connecting 5000+ apps: build cross-app workflows in ChatGPT with natural language — no automation rules to learn.

ChatGPT meets the Wolfram Alpha compute engine: complex math, statistics and scientific data with far higher precision than a plain LLM.

Anthropic's official skill: turn any content into beautifully laid-out PowerPoint decks with templates, masters and speaker notes.

Anthropic's official all-in-one PDF skill: merge, split, fill forms, sign, watermark, extract tables and OCR scanned documents.

Anthropic's official frontend design skill: web pages that shed the generic "AI look" — polished, modern interfaces ready to ship.

Anthropic's official visual design skill: posters, covers and cards produced with real design philosophy, exported straight to PNG / PDF.

Generative art with p5.js: flow fields, particle systems, seeded randomness with interactive parameter tuning — gallery-grade visuals without coding.

An animated GIF generator optimized for chat: auto-managed size and frame rate with built-in validation — one prompt to a Slack-ready GIF.

Reskin any artifact in one step: 10 preset themes (colors + fonts), or generate a new theme from your brand for consistent decks, docs and pages.

The official toolkit for complex multi-component web apps: React + Tailwind + shadcn/ui with state management and routing, deploy-ready.

The official MCP server development guide: from an API to a working MCP server in Python or TypeScript, with tool-design best practices baked in.

The official Playwright testing skill: Claude drives your local web app like a real user — clicking, typing, screenshotting — to find and reproduce bugs.

Encode your brand (palette, fonts, voice) as a skill so every document, deck and page automatically matches brand — great for teams.

The official internal communication skill: status reports, leadership updates, company newsletters, FAQs and incident reports in big-company formats.

The official doc co-authoring workflow: structured writing for PRDs, technical specs and decision docs — context transfer, iteration and reader validation.

The 80k+ star memory tool: every session is auto-captured and AI-compressed, then relevant context is injected into future sessions — your agent finally remembers you.

68k+ stars: turn any codebase into an interactive knowledge graph — explore, search and ask questions on the graph. The fastest way into an unfamiliar project.

24k+ stars, Manus-style planning: persist task plans as Markdown files that survive context loss — essential for long tasks and multi-agent work.

An open-source Chinese skill collection by renowned blogger Baoyu (JimLiu), 22k+ stars: high-quality translation, writing and content-creation skills.

A 30k-star research powerhouse: 140 ready-to-use science skills plus 100+ scientific databases across biology, chemistry, medicine and drug discovery — used by 160k+ scientists.

14k+ stars: automatically convert any documentation site, GitHub repo or PDF into a Claude Skill, with conflict detection — your private knowledge becomes an AI skill.

17k stars: an unofficial Python API + agent skill for Google NotebookLM — full programmatic access, including features the web UI doesn't expose.

A 19k-star one-stop collection: 337 skills + 30 agents + 70 commands spanning engineering, marketing, product, compliance, finance and C-level advisory.

10k+ stars: 66 specialist skills for full-stack developers that turn Claude Code into a senior pair programmer — from architecture review to test strategy.

The definitive 27k-star skill map: 1400+ curated skills from official teams (Anthropic, Google, Vercel, Stripe…) and the community — hand-picked, no AI slop.

Cloudflare's official bundle: Workers, Pages, KV, R2, D1 and AI Gateway knowledge plus wrangler deployment conventions — edge development without doc-diving.

From Vercel's engineering team: Next.js best practices, caching strategy and upgrade assistant — Claude writes Next.js the officially recommended way.

Stripe's official skill: payment integration best practices plus an SDK/API upgrade assistant — webhooks, subscriptions and checkout done the official way.

Supabase's official PostgreSQL skill: schema design, row-level security and index optimization for a safe, fast database layer.

HF's official ML workflow bundle: model training (SFT/DPO/GRPO), dataset management, evaluation, Gradio apps and in-browser inference — 13 skills covering the whole pipeline.

Official skills from elite security firm Trail of Bits: static analysis, differential review, constant-time analysis, smart-contract audits and Semgrep rule writing.

28 official Sentry skills: auto-detect your stack and wire up the SDK, fix production errors with full context, review code against Sentry data.

11 official Expo skills: native UI, routing, API routes, CI/CD and store deployment — official guidance for the whole cross-platform app lifecycle.

Firecrawl's official skill: integrate web search, scraping, structured extraction and multi-step browser flows into your app — reliable data even on anti-bot sites.

Composio's official skill: connect AI agents to 1000+ external apps (Gmail, Notion, Slack, GitHub…) with managed auth — no OAuth configuration hell.

Remotion's official skill: generate video programmatically with React — product demos, data animations and personalized video at scale. Change code, change the film.

Typefully's official skill: draft, schedule and publish X / LinkedIn / Threads / Bluesky content from a conversation — multi-platform social in one place.

7 official Binance skills: market rankings, meme-coin tracking, smart-money signals, token security audits and wallet portfolio lookups — on-chain analysis by conversation.

Brave's official search bundle: web, image, video, news and local business search plus AI-generated answers — privacy-first, real-time retrieval.

32 open-source marketing skills by expert Corey Haines: SEO, copywriting, cold email, pricing strategy, CRO and growth experiments — the full SaaS marketing stack.

OpenAI's official Codex skill: generate, remix and manage short video clips via the Sora API — text straight to cinematic footage.

OpenAI's official skill: batch-generate and edit images for your projects via the Image API — illustrations, icons and assets produced inside your dev flow.

OpenAI's official skill: complete scaffolding for ChatGPT Apps SDK projects — MCP server + widget UI end to end. Build an app that runs inside ChatGPT.

OpenAI's official skill: agents drive a real browser to navigate, fill forms and scrape — with persistent sessions for debugging complex interactions.

OpenAI's official skill: when PR checks fail, skip the log-diving — it inspects GitHub Actions failures, finds the root cause and lands the fix.

Built by this site: a work-discipline protocol that makes Opus 4.8-class models deliver Fable-5-grade quality — plan gates, task cards, adversarial self-checks and real-product verification buy single-pass quality.

A complete software-development methodology for coding agents, built from composable skills. Your agent first pins down what you're really building, drafts a spec you approve in chunks, then runs a red/green TDD plan through subagents — all triggered automatically.

A skill/plugin that makes your AI coding agent answer in terse "caveman" style — filler gone, output tokens cut by ~65%, while code, commands and error text stay byte-for-byte exact and nothing technical is lost.

Packages the workflows and quality gates senior engineers use — across defining, planning, building, testing, reviewing and shipping — into 24 reusable skills so AI coding agents follow the same production-grade practices every time.

The agent skills Matt Pocock uses every day for real engineering, not vibe coding. Deliberately small, easy to adapt and composable, and they work with any model. The core idea: run a grilling/alignment pass before the agent starts, so it doesn't build the wrong thing.

An AI agent skill that searches Reddit, X, YouTube, TikTok, HN, Polymarket and GitHub in parallel. It scores by upvotes, likes and real-money bets rather than editors, then an AI judge synthesizes one brief on what a person or topic has been doing in the last 30 days.

A set of AI skills open-sourced by Khazix — the ones he actually uses daily. They follow the open Agent Skills standard and install on Claude Code, Codex, Cursor and 40+ other agents. Each is a structured instruction set battle-tested in real projects.

Hallmark is a design skill from Together AI that refuses to produce AI-generated-looking pages. For each brief it picks a macrostructure, dresses it in one of twenty themes, and runs fifty-seven slop-test gates plus a pre-emit self-critique — so two different briefs come out looking like two distinct sites, not colour-swaps of one template.

i-have-adhd is an output-style skill for your coding agent that stops it from burying the answer. Action first, steps numbered, and every turn ends with one concrete next step — no 'Great question!', no 'Hope this helps!'. Ten rules in total, installed in one command from the Claude plugin marketplace.

Huashu Design is an HTML-native design skill: say one sentence in your agent and, in 3–30 minutes, get back a shippable launch animation, a clickable App prototype, an editable slide deck, or a print-grade infographic. Give it your brand assets and it reads your style; give it nothing and a 40-style HTML-native library (20 web + 20 PPT) plus a three-logic advisor keeps output off the AI-slop defaults. Agent-agnostic — Claude Code, Cursor, Codex and more.

book-to-skill distills any technical book (PDF/EPUB/DOCX and more) into a skill your agent loads on demand: it generates core mental models plus a chapter index, per-chapter files, a glossary, a patterns file, and a cheatsheet, with chapters loaded only when asked. Answering one question costs 24–51× fewer tokens than dumping the book into context, and answers come from the real text — no hallucination. It follows the open Agent Skills standard and works in Claude Code, GitHub Copilot CLI, and Amp.

An academic research skill pack for Claude Code, covering the full paper pipeline from research and writing through peer review, revision and finalizing. It offloads the grunt work — hunting references, formatting citations, verifying data, checking logical consistency — while trust-chain provenance and claim-audit gates curb hallucination, keeping the human in the loop rather than fully automating.

A GitHub-run community marketplace of Copilot customizations — hundreds of installable agents, instructions, skills, plugins, and workflows. Install one from the marketplace with a single command to give Copilot coding standards and task-specific abilities.

A local-first AI "second brain" for Claude Code that reads any source and files it into linked, source-cited Obsidian Markdown notes. Ships 15 skills for capture, filing, retrieval, maintenance, and visual mapping; your notes stay plain text you own, making it an open-source Notion alternative and personal knowledge management (PKM) system.

One line gives your AI agent "eyes on the internet" — read and search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu, and any web page, for free. Each platform uses primary-plus-fallback multi-backend routing, so when one access method breaks it switches automatically — no paid APIs, proxies, or logins to wrangle yourself.

Portable Agent Skills (installable like a Claude skills marketplace pack) that fix generic AI-generated UI. They give agents such as Claude Code, Codex, and Cursor stronger layout, typography, motion, and spacing taste, plus image-generation skills for design reference boards and brand kits.

An AI-powered reverse-engineering and authorized-pentest skill router for coding clients (Claude Code, Codex CLI, Cursor, Cline). When an agent hits an APK, a binary, obfuscated JS, a CTF, or a pentest target, it routes to the right methodology, checks available tools, and runs a repeatable workflow instead of guessing commands. Covers 20+ security domains with auth/scope gating before any action.

A plugin and skills marketplace for Claude Code: 94 install-on-demand plugins bundling 203 domain agents, 175 skills, and 109 slash commands. Everything is generated from one Markdown source, so Codex CLI, Cursor, OpenCode, Gemini CLI, and GitHub Copilot consume it natively too.

An open-source agentic video production system that turns your AI coding assistant into a full video studio. Describe what you want in plain language and the agent handles research, scripting, asset generation, editing and final composition. Ships 12 pipelines, 100+ tools and 700+ skill and knowledge files.

A Markdown-only skill pack that lets an agent do machine-learning research "in its sleep": from idea discovery through running experiments to writing the paper, with cross-model review loops keeping each step honest. No framework, no lock-in — runs on Claude Code, Codex, Cursor and more.

Hand a document to your AI agent and it generates a browser-editable web PPT where every page ships with an edit console—tweak anything in the browser, then export to a real, editable PPTX. Ships 12 visual themes and 1020 layout pages.

A portable agent skill that removes the signs of AI-generated writing so text reads naturally and human. Built on Wikipedia's "Signs of AI writing" guide, it detects 33 AI-isms and can even match your personal voice when rewriting.

An all-in-one skill pack for writing web fiction, covering the full pipeline for long- and short-form novels: scanning bestseller charts, deconstructing hits, writing, de-slopping AI tone, and cover art. Its method: "tropes = deterministic emotional payoff."

An open-source cybersecurity skills library for AI agents: 817 structured security skills across 29 domains (pen-testing, incident response, malware analysis, threat hunting). Every skill maps to 6 frameworks—MITRE ATT&CK, NIST CSF 2.0, ATLAS, D3FEND and more. A community project, not affiliated with Anthropic; authorized, lawful use only.

A universal SEO skill for Claude Code that runs 25 sub-skills and 18 specialist sub-agents in parallel across technical SEO, E-E-A-T content quality, Schema.org markup, GEO/AEO (AI-search optimization), local, e-commerce and international SEO. Every audit produces a prioritized, testable action plan.

Vercel's official collection of agent skills for AI coding agents, following the agentskills.io Agent Skills format. Drop it into Claude Code, Cursor or Codex and your agent audits React/Next.js performance, checks UI accessibility, and optimizes Vercel deployment cost and docs voice using Vercel Engineering's own rules.

Packages Claude Design — the design engine behind claude.ai/design — as a portable Agent Skill. Drop it into Claude Code, Cursor or Codex and produce polished UI mockups, interactive prototypes, wireframes, landing pages, dashboards and slide decks right in your editor, all as self-contained HTML that stays in your repo — no website, subscription or upload.

A plug-and-play pack of 268 AI expert agents spanning 20 departments — engineering, design, marketing, product, game, security, finance and more — each with its own persona, workflow and deliverables. Works with 18 tools including Claude Code, Cursor, Copilot and OpenClaw, and adds 53 original agents built for the Chinese market (Xiaohongshu, Douyin, WeChat, Feishu ops).

An open-source design-method engine that gives Claude Code, Codex and Cursor fixed design judgment — not one fixed aesthetic. It selects or derives the right design grammar from your references, locks project decisions into STYLESEED.md, then runs a 0–100 scored gate (nothing under 80 ships) with a render → score → revise loop to kill the tell-tale "AI-generated look."

A set of Agent Skills from kepano, Obsidian's own creator, that teach your agent to use the Obsidian CLI and its open formats — Obsidian Flavored Markdown, Bases and JSON Canvas. They follow the agentskills.io Agent Skills spec, so any SKILL.md-compatible agent (Claude Code, Codex, OpenCode) can install them to read, write and manage your vault.

A lightweight, agent-driven memory system that gives you a shared knowledge wiki across sessions, agents and devices, and automatically distills reusable Markdown skills from your history. Its core memory logic is only ~500 lines — compact enough to inspect. Installs via SKILL.md and works with Claude Code, Codex, Cursor, OpenClaw and more.

An agent skill that turns one topic into a finished Vox-style paper-collage explainer/ad video — script, collage keyframes, motion, voice-over, music and captions, all automated. Runs on the Atlas Cloud API plus local ffmpeg and works with any coding agent.

An AI video skill for Claude Code and Codex that crafts cinematic product videos with Remotion. It ships 100+ shot recipe cards, 161 motion previews and a production-ready template, with real page captures, 2.5D camera moves, beat-synced cuts and film-grade SFX.

Turns a single Claude Code session into a full game-dev studio: 49 specialized AI agents, 73 workflow skills (slash commands), and a coordination system mirroring a real studio hierarchy of directors, department leads and specialists — with 12 validation hooks and 11 path-scoped coding rules.

A Claude Code skill bundle for bug hunting and external red-team work: 82 skills, 15 slash commands, 681 disclosed-report patterns curated across 24 core vulnerability classes, plus enterprise identity and infrastructure attack matrices. Skills load automatically by topic and stay in scope.

A skill that audits and rewrites content to strip out "AI-isms." It offers detect-only, rewrite and edit-in-place modes, with a 112-word replacement table across three tiers and 60 pattern categories, plus optional voice profiles (casual/professional/technical/warm/blunt). Works with Claude Code, Cursor, OpenClaw and other agents.

Official Google Labs Agent Skills pack that pairs with the Stitch MCP server to hand UI design and component-building workflows to your AI. It follows the open Agent Skills standard, so it installs into Claude Code, Cursor, Codex and other coding agents.

A Claude skills marketplace for product managers, packing discovery, strategy, execution, launch and growth workflows into 68 skills and 42 chained commands across 9 plugins. Each skill encodes a proven PM framework and walks you step by step from idea to decision.

Distills the frameworks and methodologies buried in books, long videos, podcasts, interviews and courses into callable Agent Skills — not just summaries. Its RIA-TV++ seven-stage pipeline extracts, triple-verifies and stress-tests each unit into a triggerable, composable skill pack.

A Claude agent skill that turns cryptic terminal ASCII diagrams and pipe-and-dash tables into polished, browser-ready HTML pages or slide decks. Great for architecture diagrams, diff reviews, plan audits, data tables and project recaps.

An open-source library of 98 AI research and engineering skills. Package them into your Claude Code, Codex or Gemini agent and it becomes an AI research agent that can go from idea and literature survey to running experiments and writing the paper. Spans 23 categories, sourced from official repos and real GitHub issues.

An experimental collection of Claude plugins and skills for German legal practice, built by fully translating Anthropic's official claude-for-legal skills into German and restructuring them around German legal methodology (Anspruchsgrundlagen, review structure, case citations with date and docket number). Covers employment, corporate, insolvency, data protection, litigation and dozens more areas. For reference only, not legal advice.

Turns natural language, codebases, Terraform/Kubernetes files, or SQL DDL into editable .drawio diagrams and exports PNG, SVG, PDF, or JPG. It supports Claude Code, Cursor, Copilot, Codex, and other Agent Skills clients.

A single-file HTML presentation workflow for Claude Code, Codex, and similar agents, with Editorial Magazine and Swiss International visual systems. It includes horizontal navigation, imagery and covers, speaker notes, rehearsal, and a dual-window presenter mode.

Turns articles, copy, screenshots, product notes, subtitles, photos, or user videos into Xiaohongshu carousels, Live Photo motion cards, and paired 21:9 + 1:1 WeChat covers. It includes Editorial and Swiss content planning, layouts, and QA.

A general prompt-writing skill for Claude that detects the target tool, extracts nine intent dimensions, asks no more than three questions when critical context is missing, and selects an appropriate prompt structure. It covers coding, image, video, voice, automation, and search tools.

A provider-neutral skill for Codex, Claude Code, and other agents that designs MVPs and audits or refactors agent harnesses. It covers tool permissions, loop budgets, context and memory, orchestration, observability, evals, caching, connectors, and launch gates.

A local job-search command center for Claude Code, Codex, OpenCode, and other coding agents. It scans roles, evaluates job descriptions, builds tailored resumes, tracks applications, and supports interview prep and outreach.

Turns a theme, article, object, or reference photo into an original one-ink or two-ink editorial poster using neutral paper, halftone imagery, active negative space, and controlled typographic tension.

Runs structured multi-persona deliberation for complex decisions with an 18-member full council, focused triads, duo debates, and a quick mode. It installs for Claude Code, Codex, Gemini CLI, and OpenCode.

PostgreSQL Agent Skills maintained by Timescale, covering table design, indexes and constraints, pgvector and PostGIS, hybrid search, safe migrations, TimescaleDB hypertables, and database management.

A Claude Code workflow for premium scroll-driven landing pages. It interviews the brand, maps the customer journey, designs scroll timing, video scrubbing, horizontal rails and signature interactions, then verifies the result with scroll screenshots.

A six-mode review suite grounded in twelve classic software-engineering books. It audits pull requests, architecture, technical debt, tests, and project health, returning structured symptom, source, consequence, remedy, and health-score findings.

A diagnose-first Agent Skill for Terraform and OpenTofu modules, tests, CI/CD, state, and security. It routes work by risks such as identity churn, secret exposure, blast radius, CI drift, state corruption, and provider upgrades.

Redesigns a GitHub repository homepage as a clear, project-native visual story. It extracts the audience, value, and real proof before improving information order, Markdown, and SVG visuals, or can create only a hero, section headers, diagrams, or badges.

Turns a roughly five-second voiceover, opinion, or abstract idea into editorial halftone paper-collage B-roll. It requires approval of the visual metaphor, then the color still, before Gemini Omni Flash assembles the final animation and runs QA.

Microsoft's official Claude Code and GitHub Copilot plugin marketplace for Power Platform. It provides reusable skills, agents, and commands for Power Pages, model-driven apps, MCP Apps, Code Apps, mobile and Canvas apps, and Power Automate.

A KiCad electronics toolkit for Claude Code, Codex, and other agents. It parses and reviews schematics, PCBs, Gerbers, and reference PDFs, with SPICE simulation, EMC pre-compliance, BOM, datasheet, sourcing, and manufacturing workflows.

Provides Claude Code with 29 scripts for iOS builds, testing, and simulator automation. It favors low-token semantic navigation through the accessibility tree and supports xcodebuild, app lifecycle management, accessibility audits, visual diffs, and diagnostic snapshots.

Stores source material in an immutable raw layer and compiles it into a durable wiki. It supports incremental ingest, citation-backed queries, answer archiving, and checks for index, link, and evidence consistency.

A nine-skill WeChat Official Account pipeline covering topics, writing, review, formatting, images, assets, and publishing. Publishing defaults to the draft box and requires an explicit user request to go live.

More than one hundred reusable skills for modern Java enterprise projects, spanning requirements, architecture, planning, Maven, testing, refactoring, performance, observability, Spring Boot, and Quarkus.

Task-loaded skills derived from official Flutter, Dart, Firebase, and package documentation, covering architecture, state management, testing, Firebase, accessibility, and store-listing assets.

Forty specialist skills for the App Store and Google Play, covering keywords, metadata, competitors, creative assets, reviews, launch, paid acquisition, subscriptions, retention, analytics, and market intelligence, with a routing skill for dispatch.

Turns a local web project or hosted page into a Dock-launchable macOS app. It inspects the project, selects WebKit or a compatible fallback, creates the launcher and icon, and verifies ports, launch, warm relaunch, quit, and cleanup.

Before more building, content, or spending, identifies who pays, what outcome they buy, and which event counts as revenue, then narrows up to three candidates to one lawful, verifiable, deliverable next step closest to payment.

An adaptive learning system organized around a topic folder. It creates a syllabus and one lesson at a time, waits for reading and feedback before adapting the next lesson, and finishes with an evaluation and traceable summary.

Turns one character, creature, vehicle, weapon, or prop image into a style-consistent transparent action sequence. It locks identity and limb topology, redraws key poses, creates in-betweens, removes backgrounds, normalizes canvases, and packs spritesheets.

Fifty-five on-demand skills for Godot 4.3+, covering project structure, GDScript and C#, 2D/3D, animation, UI, physics, audio, multiplayer, mobile, XR, native extensions, testing, and optimization.

A Blender production pack for Cursor, Claude Code, Kiro, and Codex with 94 domain-loaded skills covering modeling, materials, lighting, animation, optimization, engine export, and reference matching through Blender MCP.

Seven Agent Skills-standard guides for Symfony UX that help choose and implement Stimulus, Turbo, TwigComponent, LiveComponent, UX Icons, and UX Map patterns.

A full lifecycle method for creating, improving, validating, reviewing, and packaging Agent Skills, with architecture selection, no-skill baselines, binary evidence evaluation, blind comparison, held-out gates, and bounded self-update.

Plans family-history research with the Genealogical Proof Standard, covering research questions, source priorities, the FAN principle, citations for 14 record types, evidence quality, conflict resolution, research logs, and proof arguments.

Adds deterministic secret-leak checks at Claude Code and Codex tool boundaries. It blocks sensitive-file reads, credential-like writes, and shell exfiltration patterns, with gitleaks, Git hooks, and CI as defense in depth.

When a user explicitly asks to go beyond the pragmatic answer, it maps 3–6 advanced data structures, distributed algorithms, frameworks, patterns, or infrastructure choices and ranks them by calibrated complexity, learning value, and concrete payoff conditions.

Seven provider-neutral skills for task splitting, context budgets, plan-act-reflect loops, output reconciliation, adversarial debate, proposal-only shared memory, and evidence acceptance while preserving final human authority.

An open-source social-content workspace from Zhejiang University and Peking University teams. Its 112 OpenClaw skills connect trend discovery, planning, copy, image/audio/video production, six-platform publishing, and performance review.

A creative-coding plugin for Web, Jetpack Compose, and SwiftUI. Its cast and paint pipelines route across 15 modules, propose a reviewable interaction or visual thesis before code, and then build motion, micro-interactions, and design systems.

An assess-first modernization workflow for PHP 8.1–8.5 projects. It covers PHPStan, Rector, PER-CS, DTOs, enums, readonly boundaries, Symfony, and Doctrine while keeping risky bulk rewrites behind a dry-run and review gate.

Guides coding agents to manage reversible local changes with jj in an existing Jujutsu repository while Git remains the remote, PR, CI, and audit contract. It covers operation-log recovery, noninteractive splitting, parallel workspaces, and safe handoff.

Two agent skills for editable 3D building scenes: pascal-3d handles safe connection, modeling, validation, and handoff, while furniture-fit evaluates measured footprints and collisions without overstating height, door-swing, or delivery-route evidence.

Publishes substantial implementation, investigation, review, or design results as self-contained local HTML briefings. One to six lanes progressively reveal claims, explanations, evidence, trust labels, and questions beside the relevant result.

Builds and strengthens unit, functional, architecture, E2E, integration, and mutation testing for TYPO3 extensions. It covers PHPUnit 11–13, TYPO3 12–14.3 LTS, multi-database matrices, Playwright accessibility checks, and CI diagnosis.

A maintained plugin of 35 engineering agent skills that connects requirements, planning, implementation, simplification, code review, and captured project learning into a repeatable loop across Claude Code, Codex, and other agent hosts.

Open agent plans, specs, Markdown, HTML, local diffs, and remote pull requests in a browser-based review surface, annotate or approve them, and return structured feedback directly to Claude Code, Codex, and other agents.

A 31-skill DevOps pack for Claude Code and Codex: 16 generators, 14 validators, and one Kubernetes debugger spanning Terraform, Ansible, CI/CD, containers, Kubernetes, Helm, logging, and Prometheus.

Eleven agent skills for AI short drama and motion-comic production, spanning novel analysis, project development, episodic writing, character assets, image prompts, storyboards, video prompts, generation, editing, and final review.

Give text-only agents structured visual evidence through OCR, layout, semantics, and traceable JSON descriptions, so they can analyze screenshots, interfaces, charts, and photos without guessing unseen pixels.

A local-first configuration and skills suite for Claude Code, Codex, and Gemini that constrains coding agents with hard session budgets, dangerous-command and secret-write guards, plugin audits, a sandbox entry point, and bounded PR repair loops.

Two video-analysis skills for Claude Code and Codex: video-shots measures cuts, duration, and motion with ffmpeg to build a shot-by-shot breakdown, while video-sync composites the source beside shot data that scrolls and highlights at each cut.

Use one Agent Skill and the genoffice CLI to create, read, convert, and structurally edit DOCX, XLSX, PPTX, PDF, Markdown, and HTML locally, with rendering, audits, and dry-run checks.

A pack of eleven Claude Skills for LinkedIn posts, comments, replies, DMs, profile audits, weekly planning, carousels, and repurposing, plus local Python rules for text cleanup and heuristic checks.

Four real skills and companion rules for coding agents: compress prose, reduce code through a YAGNI ladder, and recoverably trim oversized tool output on supported hosts.