Best skills & plugins for Claude Code
514 skills tracked·ranked quality-first·Updated July 2026
The short answer
The strongest Claude Code skills pair a real SKILL.md with worked examples, gotchas, and verification steps — the signals that separate a skill an agent can actually follow from a bare prompt. Skill of Skills ranks 514 Claude Code tools quality-first (structure 55%, reputation 15%, recency 15%, momentum 15%), so the highest-viability ones surface without you auditing dozens of repos.
Claude Code is where most of this directory's traffic lives, and the ecosystem has grown far past what any single awesome-list can vet. This guide ranks Claude Code skills, plugins and MCP servers by a composite quality score — structural signals like progressive disclosure, examples, and verification layers dominate, and raw GitHub stars are deliberately a minority input. Each entry carries an assessed risk level so you can weigh what a skill is allowed to do before you install it.
Top 12 ranked quality-first
| # | Skill | Tier | Quality | Risk | Platforms |
|---|---|---|---|---|---|
| 1 | affaan-m/ECC The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond. | Curated | 160 | high | Claude Code, Cursor, Codex |
| 2 | graphify AI coding assistant skill (Claude Code, Codex, OpenCode, Cursor, Gemini CLI, and more). Turn any folder of code, SQL schemas, R scripts, shell scripts, docs, papers, images, or videos into a queryable knowledge graph. App code + database schema + infrastructure in one graph. | Curated | 165 | high | Claude Code, Codex |
| 3 | openai/skills Skills Catalog for Codex | Promising | 90 | low | Claude Code |
| 4 | openai/openai-agents-python A lightweight, powerful framework for multi-agent workflows | Promising | 109 | medium | Claude Code, Codex |
| 5 | anthropics/claude-agent-sdk-python | Promising | 106 | medium | Claude Code |
| 6 | last30days-skill Claude Code skill that researches any topic across Reddit + X from the last 30 days, then writes copy-paste-ready prompts | Curated | 149 | low | Claude Code |
| 7 | anthropics/financial-services | Promising | 95 | medium | Claude Code |
| 8 | colbymchenry/codegraph Pre-indexed code knowledge graph for Claude Code, Codex, Gemini, Cursor, OpenCode, AntiGravity, Kiro, and Hermes Agent — fewer tokens, fewer tool calls, 100% local | Curated | 146 | medium | Claude Code |
| 9 | NVIDIA/SkillSpector Security scanner for AI agent skills. Detect vulnerabilities, malicious patterns, and security risks. | Curated | 140 | low | Claude Code |
| 10 | everything-claude-code Complete Claude Code configuration collection - agents, skills, hooks, commands, rules, MCPs. Battle-tested configs from an Anthropic hackathon winner. | Curated | 125 | medium | Claude Code |
| 11 | cli-printing-press Every API has a secret identity. This finds it, absorbs every feature from every competing tool, then builds the GOAT CLI — designed for AI agents first, with SQLite sync, offline search, and compound insight commands. | Curated | 169 | high | Claude Code, Codex |
| 12 | openai/openai-agents-js A lightweight, powerful framework for multi-agent workflows and voice agents | Promising | 109 | medium | Claude Code, Codex |
Showing the top 12 of 514. Scores are evidence-based quality metadata, not endorsements — review each skill's risk level and source before installing.
Frequently asked
addWhat makes a Claude Code skill high quality?
A structured SKILL.md, concrete examples, documented gotchas and edge cases, verification steps, and progressive disclosure. Skill of Skills scores these 12 structural signals and weights them at 55% of the composite — well above popularity.
addAre these skills safe to install?
Every tool carries an assessed risk level (low / medium / high / critical) based on the permissions and capabilities it requests. Review the risk level and the source repository before installing any skill — a quality score is evidence, not an endorsement.
addHow often is this ranking updated?
Continuously. An automated discovery-and-scoring pipeline refreshes the directory every few hours and re-scores metadata daily, so the ranking reflects current maintenance and momentum.