andrej-karpathy-skills
A single CLAUDE.md file of behavioral guidelines for Claude Code, derived from Andrej Karpathy's observations on common LLM coding pitfalls. It helps avoid overcomplicated code, keep changes surgical, and surface assumptions.
Flowy lists 86 plugins in the Skill Authoring category for Claude Code. authoring or organising skills, prompts and agent workflows Every listing shows what is inside before you install, who built it, and whether it is Auto-invoked, meaning a FLOW.md router fires the right skill as you prompt instead of leaving them dormant.
A single CLAUDE.md file of behavioral guidelines for Claude Code, derived from Andrej Karpathy's observations on common LLM coding pitfalls. It helps avoid overcomplicated code, keep changes surgical, and surface assumptions.
Gives a Claude Code session helpers for product and engineering work: challenging a plan's scope, judging whether an idea is worth building before any code, tracing a bug to its root cause, and reviewing recent work patterns each week.
Turns a technical book or a folder of documents into a skill your coding agent can load on demand. It pulls out frameworks, decision rules and anti-patterns, so the agent answers from the real chapter text instead of you pasting the whole book into context.
Distills colleague, relationship, or celebrity source material such as chat logs and emails into reusable person profiles, and also sets up an owner profile so a personal agent knows who it works for.
Improves Claude agent skills through a nightly offline cycle that reviews past usage, consolidates memory, and refines skills, using training ideas like batches and validation gates without changing model weights.
Autoharness, a self-learning skill layer for Claude Code, distills skills from your real sessions, updates them as you work, and prunes the ones that stop getting used. No daemon, no benchmark.
Configures AI coding agents such as Claude Code, Cursor, and Codex with persistent memory, skills, MCP servers, and personas. Includes skills for writing pull requests, collaboration comments, and design docs.
A collection of skills for Claude Code that cover analogies, blind spot scanning, book explanations, business model canvases and content cards, with support for both Chinese and English.
Manages a shared library of agent skills for Claude Code users, letting you install, update, remove, and deploy skills per agent, organize tags and presets, search, and adopt existing skills.
Builds an AI Skill that imitates a former partner from imported chat logs, social media screenshots, photos, and your own descriptions, so it speaks in their phrasing and recalls shared places. Meant for personal memories and emotional processing only.
Gives Claude Code skills for coached book reading, senior-engineer review of your git changes, creating new skills, and auditing existing ones for quality.
A meta-skill that watches your work sessions, logs patterns and user corrections, and turns them into skill updates and new skill candidates in a weekly review for your approval.
A collection of general-purpose agent skills for diagrams, documentation, development workflow and drawing tools, installable into the skills folder of Cursor, Claude Code, Codex and similar agents. Each skill is a SKILL.md that the agent calls automatically based on its trigger keywords.
A skill for creating, improving, and evaluating reusable agent skills from workflows, prompts, and scripts. It also helps with packaging, routing, tests, and install trust checks.
Helps with daily development work by refactoring bloated agent instruction files, writing API handoff docs for frontend developers, drawing architecture diagrams, and creating slash commands. Skip the repetitive setup and documentation chores.
Turns a real workflow into a tested, installable agent skill and publishes it to a team marketplace. Useful for teams that want to automate repetitive workflows with Claude Code.
A collection of Claude Code skills for step-by-step problem solving, with adversarial verification of finished work, a loop that runs tasks with parallel subagents, a domain skill builder, and a release helper for configuration changes.
A Claude skill that designs any process, idea, or problem into an ICM workspace, folder structure as agent architecture, or restructures an existing folder, repo, or vault into one.
Helps you build Claude Code extensions yourself: write SKILL.md skills, set up hooks that react to events, create MCP servers that give Claude new tools and data sources, plan projects for Claude to execute, and audit agents. Useful once you want to tailor Claude Code to your own workflow.
Gives Claude Code skills for adding Typer CLI commands, generating pytest tests, reviewing CLI code, and creating learning paths for programming tools. Aimed at Python developers building command-line tools.
Independently written prompt templates for building AI coding agents, inspired by patterns observed in Claude Code codebase. System prompts, tool prompts, agent prompts, memory management, and multi-agent coordination.
A personal Claude Code configuration bundling skills for planning with persistent markdown files, React useEffect and performance practices, coding standards, and UI design guideline reviews.
Gives your coding agent ready-made skills for browser automation, building responsive frontend interfaces, creating Codex hooks, and driving Gemini computer-use agents, so you do not have to write those workflows from scratch.
A meta-skill that builds project skills from a codebase's rules, workflows, and lessons, gathering them into a skills directory that AI coding agents such as Claude Code and Cursor can consult before each task.
Turns rough feature requests into task contracts with explicit acceptance criteria, then estimates and verifies them. Fires when you draft a spec, plan a Rust or bash change, or ask for an estimate.
A persona skill that answers programming questions in the voice and reasoning of creator Yupi (程序员鱼皮), covering learning paths, job interviews, resume tips, tech choices and AI coding. Works with Claude Code, Cursor and OpenClaw.
My agent skills that I use every day to do real engineering - not vibe coding. Developing real applications is hard. Approaches like GSD, BMAD, and Spec-Kit try to help by owning the process.
Behavioral guidelines that help Claude Code avoid common LLM coding mistakes. Use when writing, reviewing, or refactoring code to avoid overcomplication, make surgical changes, surface assumptions, and define verifiable goals.
A structured skill framework for Claude Code covering deployment, security audits, content pipelines, scheduled jobs, API integration, and workflow automation. Skills load on demand when a matching request comes up.
A set of skills that helps coding agents like Claude Code do software iterations well, from defining the real problem through to verifying results. Covers code review, architecture analysis, entity model auditing, and challenging or pressure-testing proposals.
A meta skill for Claude Code that turns a rough workflow, prompt, SOP, or old skill into a discoverable, tested skill. It covers design, trigger evaluation, format checks, README generation, and release steps.
Gives you ready-made prompts for common work such as testing, access control and sprint planning, so you can get expert-style help from an AI assistant without writing the prompt yourself.
Helps your coding agent surface what you don't yet know before work starts: hidden assumptions, blind spots in unfamiliar code or domains, and risky technical guesses. Each finding is caught early, while it is still cheap to fix.
A skill for building production TypeScript applications with Effect, covering workflows, services, layers, schemas, configuration, schedules, caches, streams and HTTP clients.
Creates new AI skills from an idea or workflow through guided interviews and pattern detection, then evaluates and exports them for Claude Code, Cursor, Windsurf and Antigravity.
Gives AI agents skills for keeping their own work reliable: checking whether stored knowledge has gone stale, watching context window health during long sessions, coordinating several agents across sessions, and turning past failures into regression tests.
Writes and checks Agent Skills for an npm package or a project you maintain, and tests each generated example against the installed version so it stays accurate. Also reviews existing skills for clear triggers and usable instructions.
Supercharge Claude Code and Codex with security hooks, reusable skills, a lightweight statusline, and workflow automation.
Turn best-practice coding advice into AI habits. Stop reciting software engineering literature to your AI agent.
Maintain Agent Skills with functional programming discipline. Functional Skill is an engineering methodology, designed for complex Skill maintenance and iteration.
Disclaimer: This repository contains proprietary source code belonging to Anthropic that was unintentionally exposed via a source map file (.js.map) included in the official @anthropic-ai/claude-code npm package.
An Agent Skill that analyzes your repository and generates the configuration files AI agents need to contribute correctly. GitHub-native, it auto-discovers your repo's context, community health, and PR review patterns without you explaining anything.
A Claude Code plugin of custom skills for engineering, Apple app design, product management, content briefs, children's book illustration, and AI architecture, with prompt engineering help for Claude models.
Khuym is a focused Codex plugin containing five independent meta-skills and one tool-automation skill. It improves how an agent understands, researches, sequences, lands, and visualizes work without imposing a repository workflow.
Build More, Architect Dreams... With the BMad Builder!
An AI collaborator persona for Claude Code (and beyond). Six layered skills: base + art director + auditor + parallel + VI + loop. Portable across 6+ AI tools.
A self-contained Claude Code plugin that carries a feature from a one-line idea to reviewed, verified, shipped code through 23 atomic, stack-agnostic skills and a TDD implementation engine, with a living decomposition roadmap above the per-feature flow and a.
Systematically clarify intent, challenge assumptions, resolve contradictions, and align goals, constraints, risks, and success criteria.
A skill that creates, evaluates, and improves other skills. Meta-level. Architecture-first skill lifecycle: design → build → test → evaluate → package. Most skill tools jump straight to "write SKILL.md."
Helps MATLAB and Simulink users with tutoring, debugging and course AI-use policies. Covers coaching through errors and failed tests, creating hands-on exercises, and writing an AI-use policy file for a course.
Universal tool to convert Claude Code skills to Gemini CLI extensions and vice versa.
Create, evaluate, and improve OpenCode skills with a bundled workflow skill and TypeScript plugin.
Context-efficient development knowledge via progressive skill discovery.
AI Agent Skills for writing idiomatic, production-quality Go code. 20 modular skills teach AI coding assistants Go best practices derived from: Google Go Style Guide Effective Go Uber Go Style Guide Go Wiki CodeReviewComments Skills are tuned following.
Blog: Learn how to build Claude Code skills Design, scaffold, build, review, evolve, and publish production-grade Claude Code skills following the Agent Skills open standard.
Three Agent Skills for turning coding-agent work into better SKILL.md files: skill-miner, mine coding-agent history, archives, memories, and repeated work to surface skill-worthy workflows with evidence.
한국어 When AI collaboration goes off-track, you redo everything. These protocols bring the decisions that set the direction to you early, often at the plan level, before a wrong one hardens into code or other downstream work, and leave each one yours to.
A small, shared skill library by builders, for builders. This repo includes skills from me and my friends. The content comes from our own practice and selected public sources. Give them a try, read along, and enjoy the craft of building. Online docs.
Agent skills and generated runtime entry points for Spec Kit.
A collection of skills for AI coding agents. Skills are packaged instructions and scripts that extend agent capabilities.
A collection of Claude skills distributed as a Claude Code plugin marketplace. Skills are primarily designed for the Claude.ai PaaS skill compute environment; mileage will vary if used as Claude Code Agent Skills.
Agent Skills for the Venice API. Skills are self-contained folders of instructions (one SKILL.md each) that an LLM agent loads on demand to work correctly against a specific surface area of the API.
Natural-Language Programming Manager, score, check, fix, and test NL artifacts across Claude Code, Codex CLI, and Antigravity. Tier-aware scoring with per-tool overlays. Part of the xiaolai plugin marketplace.
The skill skill can help you find, improve, and publish skills, install it with npx sundial-hub add skill. You'll be able to peruse this library of curated skills, plus 50,000+ more aggregated from the web.
A complete development methodology for AI agents. From idea to production. AI agents without a system are chaotic code generators.
A Claude Code plugin providing skills and comprehensive documentation for building plugins, skills, MCP servers, and other Claude Code extensions.
One command drops a tested, versioned skill into your agent. No more ad-hoc prompts. The same rigorous steps run every time. Skills are independent files.
Agent skills that give coding agents your team's context at session start, so they stop working like strangers.
A two-layer skill for writing prompts, agent directives, skill descriptions, slash commands, and anywhere else an LLM reads instructions. Works the same in Claude Code and OpenAI Codex CLI. Same SKILL.md, same trigger surface, same outcome.
Opinionated AI coding agent and dev environment automation for macOS that dominates your dev setup like cats rule the Internet. This project is named after Arthur C.
Stop typing the same long request to your coding agent every day. This Agent Skill reads your own history, finds the asks you keep repeating, and turns each one into a short command, like /sent or /project billing. It works in Claude Code, Codex and Cursor.
26 production-ready skills for AI coding agents. Install with one command, use across Claude Code, Cursor, Kiro, Windsurf, and OpenCode.
Tinder for your Claude Code skills. Swipe through your collection and delete what's wasting context, in seconds. Works with Claude Code and OpenAI Codex. 6 commands, zero dependencies. Status (v1.8, September 2026): maintenance mode.
This repository is my Pi coding agent home directory. I keep it under version control and symlink it to ~/.pi/agent so Pi loads configuration, extensions, skills, and runtime state from here.
Agent skills extracted from real work. Each one shipped something first.
AI prompt database and builder skill for Claude Code with 2,500+ curated prompts across 19 categories and 17 AI models.
CoDD: Coherence-Driven Development, Claude Code plugin for cross-artifact change impact analysis.
ostack, AI-powered engineering team for Claude Code. Fork of gstack by Garry Tan.
We are moving from "prompt as text" to skill as runtime module. Each skill acts like a lightweight, containerized application for AI work: a stable interface, opinionated workflow, and bundled resources that can be installed, versioned, tested, and reused.
Prompting is becoming one of the most important engineering skills in the AI era. Prompt Sensei helps you improve your prompts while teaching you how to improve. Prompt Sensei is a quiet, local-first prompt coach for Claude Code and Codex.
τέχνη · /ˈtɛk.niː/ · "TEK-nee" Ancient Greek for craft · skill · art: the practical know-how of making and doing, and the shared root of both technique and technology. Aristotle placed technē (knowing how) beside epistēmē (knowing that).
The toolkit for creating reusable skills for AI coding agents. SkillKit helps you build, validate, and share skills that extend how your AI agent works. Works with Claude Code, GitHub Copilot, OpenCode, and Codex. Previously known as claude-skillkit.
The canonical home for this skill is now Jakubantalik/transitions.dev. This repository is kept as a backwards-compatible mirror so the legacy install command keeps working.