Agent Tool Library

Meta-skill that discovers and verifies other skills by install count and source reputation, then installs them. The most-installed skill in the ecosystem.

A single zero-dependency file encoding Karpathy's four rules: no silent assumptions, minimal solutions first, no orthogonal changes, verify before executing. Cuts AI-driven bugs.

The official Model Context Protocol servers repository containing reference implementations for SQLite, GitHub, PostgreSQL, Google Drive, Slack, and more.

A collection of high-quality agentic skills built for real-world developers: interactive clarification, PRD generation, strict TDD, and system diagnostics.

Connect and unify data across various platforms and databases with MindsDB as a single MCP server. Connects agents to dozens of enterprise data stores.

Forces the agent to branch multiple logic paths and evaluate them before committing to complex algorithms. Reduces hallucination in hard math/logic problems.

Scaffolds and builds MCP servers so Claude can connect to new tools and data with a consistent schema.

Applies a consistent brand system (colour, type, voice) across generated deliverables.

Drafts internal communications and announcements in a consistent, professional company voice.

Full PDF lifecycle: extract, merge/split, rotation, watermarks, form filling, encryption and OCR.

Opinionated team-of-specialists framework for Claude Code (CEO, EM, Designer, QA, Security). Automates planning, diff-aware browser-based QA, and engineering retrospectives.

Connect your agents with 2,500 APIs and 8,000+ prebuilt tools. Execute complex automation workflows dynamically.

Pushes Claude to commit to a bold design direction before coding to avoid generic AI slop. The most-used design skill.

Spreadsheets with a zero-formula-error guarantee, chart generation, conditional formatting and pivot tables.

Generates custom animated GIFs for Slack and social posts.

Presentations: master layouts, speaker notes, comments, template or from-scratch.

Creates original visual art, posters and design pieces as PNG/PDF using a design-philosophy approach.

Word docs: tables of contents, headers/footers, image insertion, tracked changes and find-and-replace.

MCP server for DeepSeek AI with chat, reasoning, multi-turn sessions, function calling, thinking mode, and cost tracking.

Official Pulumi skills: ComponentResource patterns, Automation API, and migration from Terraform/CDK/CloudFormation.

Agentic generation of WebGPU compute shaders and 3D scenes. Highly optimized for modern browser engines.

Query more than 40 apps with one binary using SQL. Connects to PostgreSQL, MySQL, SQLite, and SaaS apps. Local-first and private by design.

An open registry and ecosystem for agentic tools and skills. Streamlines discovering, testing, and integrating AI agent capabilities directly into your workflow.

The canonical, source-available skill set from Anthropic (docx, pptx, xlsx, pdf, frontend-design, canvas-design, mcp-builder, skill-creator). The reference standard for how a SKILL.md is built.

The open specification and documentation for the Agent Skills standard adopted across agents.

70+ React/Next.js performance rules from Vercel Engineering across 8 categories, prioritised by impact.

Anthropic's skill for testing live web apps end-to-end and validating real user flows.

A popular curated list of MCP servers, organized by category, for discovering new tools, frameworks, and integrations for the Model Context Protocol.

Interactive Q&A that generates a complete, well-structured SKILL.md directory. The fastest way to author your own.

Production-grade engineering skills from Addy Osmani (Google Chrome): Lighthouse-style performance, Core Web Vitals, accessibility and modern best practices.

Builds self-contained interactive HTML artifacts (dashboards, tools, mini-apps).

LLM-first SEO analysis with 16 sub-skills, 10 specialist agents and 88 evidence-collector scripts. Works in Antigravity, Codex and Claude.

One local gateway and manager for all your MCP servers, shared across every AI coding tool. Features lazy discovery and stores credentials in the OS keychain.

Hypothesis-driven, isolate-and-verify debugging methodology; reported ~2.5 hrs/week saved.

Audits UI code against 100+ accessibility, performance and UX rules. Vercel's design system as a reviewable skill.

Curated index plus the connect-apps plugin for 1000+ app integrations.

Deep domain knowledge for building programmatic video with React: animation, timing, audio, captions, 3D. Widely praised on release.

A powerful image generation tool using Google's Imagen 3.0 API through MCP. Generate high-quality images from text prompts with advanced photography and artistic controls.

Teaches Claude to install and drive the Firecrawl CLI autonomously for reliable web data and scraping.

The open agent-skills CLI. 'npx skills add' installs any skill across Claude Code, Cursor, Codex and OpenCode.

Implements local-first data syncing using CRDTs. Generates schemas and sync rules for SQLite-based offline-first web apps.

Specialised guidance for writing, debugging and maintaining Playwright tests (E2E, component, API, visual, accessibility).

Plugin-based MCP server and Chrome extension that gives AI agents access to web applications through the user's authenticated browser session.

The biggest community framework. Runs a full agentic dev workflow: brainstorm to spec to plan to subagent execution to review to merge.

Enforces strict RED-GREEN-REFACTOR; deletes any code written before a failing test exists. Part of Superpowers.

Refines rough ideas via Socratic questioning and saves a design doc before any build. The front of the Superpowers workflow.

20 SEO & GEO skills for 35+ agents: keyword research, content writing, technical audits and rank tracking (CORE-EEAT + CITE).

Universal SEO skill: 25 sub-skills + 18 sub-agents covering technical SEO, E-E-A-T, schema, GEO/AEO, backlinks, local SEO and PDF/Excel reporting.

Local-first knowledge graph for developers. Watches project files, extracts entities and relationships via LLMs, and builds a queryable knowledge graph.

Curated index of 1000+ skills across Claude Code, Codex, Cursor and Gemini CLI. Best starting point.

Memory/context engine topping three memory benchmarks. Tracks user facts over time, resolves contradictions and auto-forgets expired info.

Structures agent work as plan files on disk so long tasks stay on-track. One of the most-starred individual skills.

Headless browser automation for agents with reliable ref-based targeting, parallel sessions and cloud-browser support.

Standardises 1000+ external APIs into typed, callable tools with OAuth lifecycle management and observability.

A widely cited collection of production-ready subagents and skills for Claude Code.

Model-invoked browser automation: Claude autonomously writes and runs Playwright automation for testing and validation.

Professional-grade static analysis and vulnerability detection (Slither, Echidna) from a top security research firm. The gold standard for security skills.

A design-reasoning engine: 50+ styles, 161 palettes, 57 font pairings, 99 UX guidelines and a full design-system generator.

Writes clean, Conventional-Commits-style messages from your staged git changes.

A DevOps skills marketplace: Terraform/Terragrunt IaC, Kubernetes troubleshooting, CI/CD and incident response.

Chat with your database or datalake (SQL, CSV, parquet). Makes data analysis conversational with LLMs + RAG.

Isolated E2B cloud sandbox to plan, build, host and test full-stack apps without touching local or production.

Senior-engineer review of git changes; flags SOLID violations, security risks (XSS, injection, SSRF), N+1 queries and dead code, tagged P0-P3. Review-only by default.