Claude Code Skills Ranking
Multi-dimensional scoring, updated weekly — find the best Skills to install
Last updated: 2026-10-09Oathe: https://oathe.ai/report/mattpocock/skills
TypeScript master skills: type inference, refactoring, best practices
“Strong instructions and safety boundaries, but 37 generic engineering skills heavily overlap Shareuhack native planning, TDD, review, and debugging; do not adopt the pack wholesale.”
Oathe: https://oathe.ai/report/obra/superpowers
Comprehensive agentic dev skills: TDD, debugging, code review, planning
“Instructions are strong, but subagent and Git authority is broad; use only in isolated, supervised development workflows.”
Oathe: https://oathe.ai/report/addyosmani/agent-skills
Web quality skills: performance, accessibility, code quality analysis
“Quality and security guidance are strong, but planning, TDD, review, performance, accessibility, and Git workflows substantially overlap Shareuhack capabilities.”
ClawSecure: https://www.clawsecure.ai/report/919ad1e94278fcc3
Anthropic's official skills repo: creative, technical, enterprise, and document skills
“Mature instructions and validation, but a broad document/browser subprocess surface and no root license or SECURITY.md.”
Marketing skills: SEO copywriting, social media posts, marketing strategy
“High-quality marketing instructions with no default publishing, but heavy overlap with Shareuhack content and growth skills.”
Oplane: https://www.oplane.io/blog/karpathy-skills-security-blind-spot
Andrej Karpathy's LLM coding principles and best practices
“Concise and verifiable, but preserving existing style and avoiding adjacent fixes can suppress nearby security corrections, while the guidance overlaps native engineering rules.”
Oathe: https://oathe.ai/report/Leonxlnx/taste-skill
Anti-slop frontend skills: gives AI agents good taste and stops generic boring UI generation. Works with Claude Code, Cursor, Codex, Gemini CLI
“Static risk is modest, but failure and verification contracts are weak and design overlap is high.”
Academic Research Skills for Claude Code: research → write → review → revise → finalize — full 10-stage academic pipeline
“Rigorous research workflow, but auto permissions, external sources, and cross-model dependencies do not fit the existing content pipeline directly.”
NLPM: https://nlpm.com/reports/nexu-io-open-design
Open-source design skill: rapid prototyping, design systems, component library generation
“Capable but too large and permission-heavy, with substantial overlap with existing design and Figma capabilities.”
Turn any codebase into an interactive knowledge graph: multi-agent analysis pipeline, visual exploration, works with Claude Code, Codex, Cursor, Copilot, Gemini CLI
“Knowledge-graph outputs are clear, but multiple agents, parsers, and local services add authority and operational burden.”
Skillstore: https://skillstore.io/skills/kepano-obsidian-bases/audits/8
Obsidian knowledge management: note organization, knowledge graphs, Markdown workflows
“Useful, verifiable Obsidian formats; isolate Defuddle network input and its global npm install.”
Manifold Security: https://manifest.manifold.security/skills/skills.sh/mvanhorn/last30days-skill/last30days
AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web — then synthesizes a grounded summary
“Research output and verification are clear, but cookies, API keys, and multi-source network input do not suit unattended operation.”
Agent harness performance optimization: skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond
“Broad coverage without a consistent package-level contract; installer and hook authority do not fit Shareuhack cron.”
Oathe: https://oathe.ai/report/OthmanAdi/planning-with-files
Project planning aid: task decomposition, progress tracking, file-based planning
“Well specified, but global hooks and cross-session reads create sensitive-data and context risks.”
Anthropic official document skills: PDF, Word, PowerPoint, Excel
“The source path is stale, so the current document skills cannot be statically reviewed.”
Anthropic official frontend design skill for rapid UI prototyping
“The source path is stale, so the current frontend skill cannot be statically reviewed.”
Oathe: https://oathe.ai/report/nextlevelbuilder/ui-ux-pro-max-skill
AI design intelligence skill: 50+ UI styles, 161 color palettes, 57 font pairings across 12 tech stacks. Works with Claude Code, Cursor, Windsurf
“Instructions are structured, but current static review and published CVEs expose broad shell, network, home-directory, and installer risk; do not install pending further hardening.”
Pre-indexed code knowledge graph for Claude Code, Codex, Gemini, Cursor — fewer tokens, fewer tool calls, 100% local
“The cross-platform local index is useful, but no matching end-user SKILL.md exists and installer/subprocess risk is high.”
Anthropic official MCP server builder with best-practice scaffolding
“The source path is stale, so the current MCP builder cannot be statically reviewed.”
Oathe: https://oathe.ai/report/juliusbrussee/caveman
Claude Code skill that makes agent talk like caveman — cutting 65% of output tokens while keeping full technical accuracy
“Instructions and rollback docs are detailed, but hooks, installers, default telemetry, credential-bearing proxy flows, and an unresolved high-risk scan issue make unattended use unsuitable.”
A meta-skill that designs domain-specific agent teams, defines specialized agents, and generates the skills they use
“Clear agent and skill output contracts, but tied to Claude teams/Opus and overlapping existing governance skills.”
MrTechKing current-commit static audit: https://www.mrtechking.com/trail-of-bits-skills-github/
Trail of Bits security skills: vulnerability scanning, code analysis, threat modeling
“Strong security instructions, but broad tooling and install authority make it unsuitable for unattended content cron use.”
Official Compound Engineering plugin for Claude Code, Codex, Cursor — knowledge compounds across sessions with brainstorm/plan/work/review/compound workflow
“Mature workflows and verification, but GitHub publishing authority and overlap with the existing agent fleet are too broad.”
817 structured cybersecurity skills for AI agents — mapped to MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND, NIST AI RMF, MITRE F3. 29 security domains, works with Claude Code and 20+ platforms
“Structurally consistent, but operational security authority is too broad and lacks a package-wide authorization boundary.”
1000+ curated production-ready Claude Skills and Plugins for Claude Code, Codex, Cursor, Gemini CLI, and more
“Large-scale remote automations lack consistent permission, failure, and rollback contracts; do not adopt directly.”
1000+ curated official and community agent skills from Anthropic, Vercel, Stripe, and more
“This is an unpinned external link catalog, not a verifiable skill package.”
Autonomous game development for Godot, Bevy, and Babylon.js with Claude Code and Codex
“Detailed asset-generation contract, but paid APIs, engine subprocesses, and a publish script do not fit content-platform cron.”
NLPM: https://github.com/xiaolai/nlpm/blob/main/auditor/audits/alirezarezvani-claude-skills.md
337+ production-ready Claude Code skills and agent plugins covering engineering, marketing, compliance, C-level advisory for 12 AI coding tools
“Instruction structure is generally strong, but user-controlled commands still reach shell=True paths; keep unreviewed until those high-risk execution paths are contained.”
Installable library of 1,500+ agentic skills for Claude Code, Cursor, Codex CLI, Gemini CLI, and more. Includes NPX installer, bundles, workflows, and official/community skill collections.
“High-risk aggregate: inconsistent content plus permission-bypass and broad execution instructions.”
Curated list of Claude Skills across 11 categories: documents, dev tools, data, research, writing, learning, media, health, security, and more
“An external skill link list without source pinning, security checks, or a verifiable execution contract.”
Curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows — particularly Claude Code
“An external link list only, with no executable or verifiable skill contract.”
FAQ
How are Claude Code Skills ranked?
We score across 6 factors: Adoption (GitHub Stars + weekly growth), Community (HN, Reddit, Twitter multi-platform signals), Quality (hands-on testing), Security (audit status + risk analysis), Credibility (official vs community), and Ecosystem (category breadth + maintenance).
What is the difference between community and official skills?
Official skills are maintained by the official team; community skills come from open-source authors and vary in quality and risk. “Official” does not automatically mean independently security-audited—use each record’s review status, evidence, and requested permissions.
How often is the ranking updated?
GitHub data (stars, forks, last push date) is updated weekly via API. Quality scores and security status are updated manually when we re-test. Skills not tested in over 45 days show a staleness warning.
How do I install Claude Code Skills?
Run the /install-skill command in your Claude Code terminal with the GitHub URL. See our complete Skills guide for detailed steps.
Rankings are generated automatically from public data, updated weekly. All third-party skills are not developed by Shareuhack. Review requested permissions before installing any skill. Rankings are for reference only — use at your own risk.