Top 100 Claude Code Skills In The Wild

Anthropic's official skill for creating, evaluating, and optimising other skills. Underpins almost every skill in the wild — used to scaffold, test triggering, and tune descriptions.

Super Master Agent for the 1G 2026 goal — Apps + Backoffice + Context. Central brain routing across ~50 abc-* sub-skills and tracking progress through Prodgame.

Master Agent for the Cleverfox AIBO platform — the AI-first backoffice that drives personas, assets, media, and workflows. Routes to all per-page and codebase agents.

Video A Day — full daily YouTube-video pipeline from NotebookLM cinematic to ffmpeg assembly. Mat's flagship media-automation skill with `vidai app`, sheet sync, and uploader.

Generate one fresh Top-100 list per day, cross-checked against ~810 historic lists in MongoDB AIDB.DAILY. Powers the LAD gallery on localhost:24205.

Daily 5-stream context digest — personal, project, industry, AI, world. Builds 365 structured pages per year. /pad-research is the legacy keyword variant.

Master agent for everything audio — DJ recordings, TTS, Whisper transcripts, AAD pipeline, and a Next.js gallery on port 24202. Auto-launches the app every invocation.

YouTube Partner Program watch-hour automation for Siems Production. Daily 10-minute video drops, top channel analysis, generated-video inventory.

Generates the AGENTS.md standard file for repos — the agent equivalent of CLAUDE.md. Inspired by agents.md and adopted across the open-source agent ecosystem.

Brainstorm → spec → chunked execution with subagents. Inspired by superpowers and get-shit-done. Structured replacement for ad-hoc 'just code it' prompts.

Design and build rich, structured UI responses — cards, forms, dashboards, data visualisations. Inspired by Google's A2UI proposal.

Knowledge-graph and memory management for agents. Inspired by Cognee — builds RAG pipelines and persistent agent memory across sessions.

Deep research for development decisions — web search, repo analysis, library comparison, ADR-shaped writeups. Inspired by Perplexica and autoresearch.

Create distinctive, production-grade frontend interfaces. Anthropic-published skill that avoids generic AI aesthetics — gradients, drop-shadows, glass-morphism.

Image generation wrapper around Gemini CLI's Nano Banana model. The default 'generate any image' skill — blog featured images, thumbnails, diagrams, illustrations.

Turn any URL, project, or idea into a single-file React JSX artifact using the ABC app shell — sticky header, collapsible nav + context panel, three switchable views.

Cleverfox-stack-aware system architect that turns intake + sector cards into an ADR + data_model + api_spec triad. Decision-document-shaped output, not Mermaid slop.

Turn any client brief into a full proposal pack — markdown master, Cleverfox-styled HTML, interactive JSX prototype, print PDF, and follow-up email — in one run.

Take any concept and produce a 13-slide pitch JSX + AIBO page + Remotion video + outreach email + LinkedIn post + tweet. Full multi-format content pack from a single idea.

Editorial-quality SVG diagrams (flowcharts, quadrants, pyramids, timelines) assembled into per-topic repo pages. Brand-aligned alternative to Mermaid slop.

Turn any raw client input — voice memo, transcript, scribbled bullets — into a structured Project Card that every downstream ABC skill consumes.

Classify an ABC Project Card into a sector vertical (charity, care, property, recruitment…) and attach the matching pack: jargon, KPIs, persona patterns, compliance.

Claude Code Cheat — a 4-hour, 10-POM crash course for teams adopting Claude Code. £500 product built from real production experience shipping 13 sites + bo28 + media pipelines.

Codebase cleanup pass — lint, unused imports/deps, theme consistency, file organisation. Applied across ABC repos as a pre-PR routine.

Build the ABC GWS cockpit — multi-tenant Workspace command runner with Fox personas, recipe orchestration, and live audit streaming. 5 sequenced CC prompts.

Next.js 16 App Router conventions for ABC — route groups, params as Promise, API routes, config patterns. Reference for every new ABC app.

Tailwind v4 + shadcn/ui (base-nova) for ABC. CSS import order, theme tokens, dark mode, CVA pattern, fox-orange accent.

Manage shadcn components — add, search, fix, debug, style, compose. Project context, component docs, --preset codes, components.json wiring.

ABC design system — shadcn-aligned CSS variables, primitives, brand tokens. Canonical token set read before writing the first line of styled JSX.

Whole-site UX audit across mobile/tablet/desktop — responsiveness, a11y, interaction patterns, visual consistency. Updates design system + storybook + backlog.

Master Agent for the CONTEXT pillar (~/context-2026) — Mat's private 1G planning, world context, strategic reference. PRIVATE; never exposes contents externally.

Social Media Manager — given any URL, generates 6 assets: blog post, 13-slide pitch JSX, AIBO page, email, LinkedIn, tweet. Supports `--type` for single-asset runs.

Scaffold AIBO-specific agents — codebase agents, org agents, task agents. Encodes Cleverfox's per-codebase / per-org / per-task layering.

Scaffold a new Next.js app following AIBO conventions — Next.js 15 App Router, Supabase, Tailwind, shadcn/ui, OrgProvider, ToastProvider.

Turn the contents of briefs/<vertical>/ into a working /<vertical>/brief page in the AIBO app. Idempotent — re-runs update rather than duplicate.

Scaffold a vertical-specific diagrams page in the AIBO app. Mirrors the platform-wide /diagrams page but scoped to one vertical, with SVG conventions and theme tokens.

Generate a landscape Cleverfox-branded PDF of AIBO documentation. Orange/black/white branding, intended for client and stakeholder distribution.

Create and maintain an Obsidian vault mirroring AIBO docs, backlog, team notes, and session history. Markdown-compatible knowledge layer.

Coach + guide navigator + quiz runner + app launcher for the Claude Managed Agents harness (Anthropic's launched-April-2026 product). Study-friendly skill in ~/APPS/ama.

Exam coach + study-guide navigator + quiz runner for the CCA cert. Lives in ~/APPS/cac with domain quizzes and resource indexes.

Master agent for the Onyx-based enterprise AI platform at ~/CF/onyx. Setup, dev ops, debugging, migrations, service health — all via /aip.

Legacy keyword-research variant of /pad — NLM web research → blog post + portrait infographic + slide deck → HTML/PDF → GSheet log.

YPP watch-hour growth automation. Daily 10-minute YouTube video drops, top viewed channel analysis, generated-video inventory tracking.

YT-PIPELINE — one command full VAD26 workflow: research → notebook → 5 production assets (audio, video, deck, infographic L+P) → metadata → report.

NotebookLM CLI skill (notebooklm-py package). Manage notebooks, add sources, generate assets, ask questions. Used standalone or as part of /yt.

Turn any codebase/repo/URL/screenshot into a 24-second showcase MP4 — analyse → boot locally → Playwright capture → Claude script → Remotion render.

Monitor long-running lab processes — build pipelines, data imports, AI generation jobs. Tracks progress of background tasks across the lab.

Full QA lifecycle — all tests, all features, all pages. Debug failures, add bugs to backlog, fix issues, update status. End-to-end /qa.

Vercel operations — sync env vars from .env.local, manage deployments, check status, promote, rollback, configure project settings.

GitHub operations via gh CLI — starred repos, repo analysis, issues, PRs, releases, integration discovery.

Release management — version bumping, changelog generation, release notes, tag creation. Standard release pipeline for VIDAI codebases.

Validate .env files, check required variables, compare environments, generate templates. Used pre-deploy across every ABC app.

Application performance analysis — bundle size, build times, component complexity, database query patterns. Identifies and fixes bottlenecks.

Anthropic-published skill — build, debug, optimise Claude API / Anthropic SDK apps. Always includes prompt caching; handles version migrations (4.5→4.6→4.7).

Built-in skill — initialise a new CLAUDE.md file with codebase documentation. Shipped with Claude Code; usually the first skill any new repo invokes.

Built-in skill — review a pull request. Reads the diff, runs checks, leaves structured comments mirroring senior-engineer review patterns.

Built-in skill — complete a security review of pending changes on the current branch. OWASP-aligned, focuses on data flow and trust boundaries.

Run a prompt or slash command on a recurring interval. /loop 5m /foo, or omit interval to let the model self-pace. Powers polling and watch-style workflows.

Create, update, list, or run scheduled remote agents on a cron schedule or once at a specific time. Routines for autonomous tasks.

Import data from external sources — JSON files, CSV, API responses, other AgentAI instances. The 'load this data' entry point.

Push the current git branch to origin. Pre-push summary (branch, last commit, status) then git push. Universally invoked.

Review changed code for reuse, quality, and efficiency, then fix any issues found. Anthropic-published — refuses to add things, only removes.

Configure the Claude Code harness via settings.json — hooks, permissions, env vars. Anthropic-published; required for any 'whenever X happens' automation.

Customise keyboard shortcuts, rebind keys, add chord bindings, modify ~/.claude/keybindings.json. Anthropic-published.

Scan transcripts for common read-only Bash and MCP calls, then add a prioritised allowlist to .claude/settings.json to reduce permission prompts.

Reformat markdown files — remove blank lines within sections, normalise spacing around headings. Tiny, idempotent housekeeping skill.

Manage persona lifecycle — create, version, clone, compare, bulk operations. Central to AIBO-style multi-character agent ecosystems.

Review the whole site UX flow — mobile, tablet, desktop responsiveness, a11y, interaction patterns. Updates design-system and storybook pages.

Open-source meta-skill that adds 'planning, focus, ownership' subagent patterns. One of the most-forked skills in the wild.

Spec-driven skill encoding the 'brainstorm → spec → execute' loop. Influential prior art that the official /spec skill draws from.

Knowledge-graph + agent memory framework. Used directly as a skill via /knowledge — manages what agents know and remember across sessions.

Open-source Perplexity-like research assistant exposed as a skill. Foundation for the /research family that competes with WebSearch.

Autonomous multi-step research framework — surfaces, synthesises, cites. Powers the deep-research mode in several skills.

Google's Agent-to-User-Interface proposal. The conceptual basis for /rich-ui — structured cards, forms, dashboards rather than plain text.

Community skill wrapping the MCP Inspector for debugging in-development MCP servers. Pairs with the /mcp catalogue.

Skill that enumerates available MCP servers in the current installation and links to their docs. Useful as a 'what tools do I have' index.

Encapsulates branching, rebasing, and PR-prep patterns. One of the earliest community skills shared in the Claude Code Discord.

Runs the project's test suite, parses failures, surfaces them as task list items. Common starting point for new Claude Code adopters.

One-command production deploy with branch + commit + SBOM check. Common in CI/CD-heavy teams that want a single agent entry point.

dbt project navigator — run models, test, docs, lineage. Common in analytics-engineering teams adopting Claude Code.

Build Stripe integrations from product specs. Encodes Stripe API patterns, webhook handling, and test-mode/live-mode setup.

Supabase project scaffolder with RLS, auth, and edge-function patterns. Often paired with abc-nextjs or aibo-app-creator.

Fly.io deployment helper — fly.toml generation, scaling, secrets, regions. Popular in indie-dev community.

Vercel project scaffolder distinct from vidai-vercel — covers Next.js, Edge Config, KV, Postgres, and AI SDK patterns.

Mat's Obsidian-vault helper — feeds notes from CC sessions into a vault that supports daily, weekly, and topic-based views.

Cross-vault semantic search using a local embedding model. Used heavily in /aibo-vault and ABC ccc-vault workflows.

Encapsulates common Puppeteer / Playwright patterns — login flows, scraping with pagination, file download verification.

Project-management skill on top of the Linear MCP — sprints, milestones, dependency mapping, async standups. Used in cross-functional teams.

Publish output from any skill into a Notion page or database. Used as the 'persist this somewhere shareable' step in many pipelines.

Publish skill outputs into an Obsidian vault. Equivalent to notion-publish but for the local/Obsidian-first crowd.

Programmatic video rendering with Remotion. Backs /abc-social and /vidai-showcase for the actual MP4 output step.

Text-to-speech wrapper for ElevenLabs. Sits underneath /audio and /vidai-yt as the narration backbone.

Local Whisper transcription with VTT output. Used by /audio transcribe and the AAD pipeline; supports diarisation.

Polls open PRs and surfaces stalled ones, CI failures, and review-time anomalies. Often paired with /loop for continuous oversight.

Drives an on-call response — pull observability data, correlate logs, draft a postmortem skeleton. Tight pairing with Sentry MCP.

Generate API and project documentation from source code. Mintlify-flavoured output; often paired with /mintlify-publish.

Internationalisation pipeline — extract strings, sync translations, validate locale fallback. Crucial for the 13-site ABC ecosystem.

Enforce performance budgets in CI — bundle size deltas, LCP/CLS thresholds, query plan regressions. Built on top of vidai-perf patterns.

Browse and search available skills locally. The on-disk counterpart to mcpservers.org for skill discovery; usually paired with skill-creator.

The umbrella orchestrator that fronts the entire ABC family. Routes by intent across 50+ abc-* sub-skills. Most-typed command in Mat's workflow.