BehiSecc/awesome-claude-skills — Summary

A curated awesome-list of Claude Skills organized into 14 categories with no original skill files — the entire value is a navigable index of external skill repositories. The list covers document skills (DOCX/PDF/PPTX), development tools (TDD, debugging, Azure DevOps, AWS CDK), data/analysis (SQL, Kaggle, crypto), scientific/research tools, writing/research, learning, media, health/life sciences, collaboration, security/web testing, and utility/automation. It explicitly cross-references the superpowers skills, Anthropic official skills, and dozens of community contributions. The repo contains a single README.md — no code, no agents, no SKILL.md files, no hooks, no config.

differs_from_seeds: This is a pure Archetype 4 (markdown scaffold, zero primitives) but at the meta-level — not even a scaffold for agent behavior, just a navigable index. Unlike claude-conductor (Archetype 4, which provides templates the agent reads actively), this is a human-navigable reference. The closest functional analog is a curated package registry rather than any framework seed.

Status: Tier B — minimal analysis possible (README-only repo). Content is externally valuable as a discovery surface but offers no original engineering.

BehiSecc/awesome-claude-skills — Overview

Origin

Created by BehiSecc as an "awesome list" aggregator for Claude Skills. The author also created VibeSec-Skill (referenced in the README tip: "If you use Claude to build web applications, do yourself a favor and use VibeSec-Skill to avoid getting hacked.").

Content

The README is a categorized table of contents linking to external Claude skill repositories. Categories:

1. Document Skills (DOCX, PDF, PPTX, XLSX, RevealJS)
2. Development & Code Tools (TDD, debugging, Azure DevOps, AWS, Claude Code terminal title, Charles Proxy, debug-skill, plugin-authoring, Playwright, Unity agent skills, etc.)
3. Data & Analysis (CSV, NotebookLM, PostgreSQL, MySQL, SQL Server, Kaggle, DeFi, Twitter scraper, crowdcast simulation, CoinPaprika, DEX data, GitHub stars)
4. Scientific & Research Tools (125+ scientific skills, materials simulation, Gemini Deep Research, Manus AI, OpenAlex paper search)
5. Writing & Research (article extraction, content writing, internal comms, brainstorming, genealogy)
6. Learning & Knowledge
7. Media & Content
8. Health & Life Sciences
9. Collaboration & Project Management
10. Security & Web Testing
11. Utility & Automation
12. Articles & Blog Posts
13. Collections (links to other curated lists)
14. Contribution guidelines

Notable Items Listed

• claude-starter (raintree-technology): 40 auto-activating skills, TOON format for 30-60% token savings
• jeffallan/claude-skills: 65 full-stack development skills
• hashicorp/agent-skills: HashiCorp official Terraform skills
• spartan-ai-toolkit: Engineering workflow with TDD enforcement
• aws-skills: AWS CDK with cost optimization MCP
• agnix: Linter for AI agent configurations (156 rules, LSP server)

Philosophy

An "awesome list" with no stated philosophy. The tip at the top (VibeSec-Skill) suggests a security focus by the author.

BehiSecc/awesome-claude-skills — Architecture

Distribution

• GitHub: BehiSecc/awesome-claude-skills
• Install: None — this is a browsable index, not an installable package
• License: None specified
• Primary language: None (README.md only)

Structure

awesome-claude-skills/
└── README.md   # Single file — the entire repository

Content Organization

The README is organized as a numbered, categorized link list. Each entry follows the format:

- [skill-name](github-url) - One-line description.

Some entries have extended descriptions (e.g., claude-starter describes its full feature set).

No Code, No Primitives

This repository ships:

• 0 SKILL.md files
• 0 agents
• 0 hooks
• 0 scripts
• 0 MCP servers
• 0 commands

All content is links to external repositories.

GitHub Stars

9,246 stars — second-highest in this batch (after muratcankoylan-skills at 16,030). High star count reflects the value of aggregated discovery rather than original engineering.

BehiSecc/awesome-claude-skills — Components

Original Components

None.

Indexed Components (external links)

Document Skills (5 entries)

• docx, pdf, pptx, xlsx (Anthropic official), revealjs-skill

Development & Code Tools (20+ entries)

• web-artifacts-builder, test-driven-development (superpowers), using-git-worktrees (superpowers), finishing-a-development-branch (superpowers), pypict-claude-skill, aws-skills, claude-starter (40 auto-activating skills), move-code-quality-skill, oiloil-ui-ux-guide, claude-code-terminal-title, charles-proxy-extract, debug-skill, plugin-authoring, azure-devops, claude-skills (jeffallan — 65 skills), jules, hashicorp-agent-skills, agnix (156-rule linter), email-html-mjml, agent-cards-skill, Design Auditor, lightning-factory-explainer, lightning-channel-factories, lightning-architecture-review, Playwright Skill, Emblem AI Agent Wallet, unity-agent-skills, spartan-ai-toolkit

Data & Analysis (15+ entries)

• csv-data-summarizer, notebooklm, postgres, mysql, mssql, kaggle-skill, octav-api-skill, elicitation, recommendations, x-twitter-scraper, claude-ecom, chainaware-behavioral-prediction, crowdcast, coinpaprika-api, dexpaprika-api, gh-star-history

Scientific & Research (5 entries)

• claude-scientific-skills (K-Dense, 125+), materials-simulation-skills, deep-research, manus, paper-search

Writing & Research (5+ entries)

• article-extractor, content-research-writer, internal-comms, brainstorming (superpowers), family-history-research

(Additional categories not fully enumerated in examined excerpt)

Cross-References Discovered

The list explicitly references:

• obra/superpowers (test-driven-development, using-git-worktrees, finishing-a-development-branch, brainstorming)
• anthropics/skills (docx, pdf, pptx, xlsx, web-artifacts-builder, internal-comms)
• K-Dense-AI/claude-scientific-skills (claude-skills-mcp's data source)
• BehiSecc/VibeSec-Skill (author's own security skill)

BehiSecc/awesome-claude-skills — Uniqueness

differs_from_seeds

This repo does not implement any of the 5 seed archetypes — it is a meta-resource that indexes implementations from those archetypes. At 9,246 stars it is a discovery surface, not a framework. The closest seed is claude-conductor (Archetype 4: markdown scaffold, zero primitives), but even claude-conductor provides active agent-read templates, while this is a purely human-navigable index.

Value

High discovery value for researchers finding the Claude Skills ecosystem, but zero engineering value as a framework. The star count reflects community appreciation for aggregation rather than the quality of original engineering.

Notable Discovery Value

This README is one of the most comprehensive indexes of Claude Skills in existence and explicitly cross-references:

• obra/superpowers (seed framework, 207,067 stars)
• anthropics/skills (Anthropic official)
• K-Dense-AI/claude-scientific-skills (directly indexes claude-skills-mcp's data source)

The crowdcast entry is notable: "Multi-agent social simulation as a Claude Code skill. Spawns dozens of AI agents that argue, post, and react on a simulated platform... Zero dependencies."

The agnix entry (156-rule AI agent config linter with LSP server) is the most architecturally interesting tool indexed here — a static analysis tool for skill/agent config files that no seed implements.

Observable Failure Modes

1. Link rot: External repositories may move, rename, or disappear. The README has no mechanism to detect broken links.
2. Curation bias: Curated manually; no automated quality gate for included skills.
3. No versioning: Skills evolve; the index doesn't track which version of each skill was reviewed.