swe-re-expert

The Full Skill Library: At-a-Glance Overview

Agentic Awesome Skills ships an installable catalog of agent skills in skills/<skill-id>/SKILL.md. This page is a curated high-level map of the whole library: how skills are organized, the main capability areas and their approximate sizes, the risk labels you will encounter, and the authoritative surfaces to browse when you need the complete, current list.

The authoritative, machine-generated catalog is CATALOG.md. This overview is a hand-maintained index; it can lag the catalog after a release, so treat CATALOG.md and skills_index.json as the source of truth for counts and exact skill lists.

How the library is organized

Every skill directory contains a SKILL.md whose frontmatter drives discovery:

Field	Purpose
`name`	Canonical skill id, e.g. `email-issue-fixer`
`description`	Short purpose and trigger summary shown in search surfaces
`category`	Thin grouping used by UI/search (see notes below)
`tags`	Free-form keywords for search
`risk`	One of `none`, `safe`, `critical`, `offensive`, or `unknown`
`source` / `source_type`	Provenance of the skill content
`date_added`	When the skill entered the registry

The full registry is exposed through three consistent surfaces:

CATALOG.md — human-readable catalog grouped by category.
skills_index.json — the legacy compatibility manifest (documented in discovery-manifest.md).
data/catalog.json — structured catalog used by the catalog app and AAS Core.

The category field is loose and historically fragmented (hundreds of skills are unclassified). Use tags, full-text search, and the AAS Core search_skills tool instead of relying on the category field as a taxonomy.

Capability areas

The table below groups the real catalog data into editorial capability areas so you can find the region of the library that matches your task. Counts are approximate and change as the catalog evolves.

Capability area	Approx. skills	What you will find
Development & code quality	480	General, frontend, backend, mobile and fullstack development; language-specific guidance; code review and quality patterns
Other & uncategorized	330	Small, niche, or legacy skills without a clean category label
Cloud & infrastructure	215	Cloud platforms, DevOps, databases, operations, and ML infrastructure
AI, agents & ML	205	Agent workflows, orchestration, evaluation, prompt engineering, and voice agents
Automation, productivity & workflow	190	Task automation, workflows, browser automation, productivity, and collaboration
Content, writing & creative	160	Writing, design, graphics, media, video, and presentation skills
Business, product & research	160	Strategy, planning, product, project management, legal, finance, research, education, and health
Security & pentesting	85	Security engineering, offensive testing, and defensive hardening
Marketing, growth & SEO	70	Campaigns, analytics, SEO, growth, and app-store optimization
Data & document processing	65	Data science, engineering, spreadsheets, documents, and knowledge management
Testing & reliability	55	Test authoring, automation, reliability, and quality engineering
Meta, authoring & special	35	Meta skills, skill authoring, and general-purpose helper skills
API & integrations	30	Connectors and automation built around third-party APIs
Gaming & blockchain	25	Game development, blockchain, and cryptocurrency

Risk labels

Every skill carries a risk: label in its frontmatter:

Risk	Meaning
`none`	No elevated risk; plain productivity guidance
`safe`	Generally safe; local or low-impact operations
`critical`	Privileged, destructive, or high-impact operations (code execution, deployments, pipelines)
`offensive`	Offensive security / penetration-testing content; review the security guardrails before use
`unknown`	Legacy or unclassified content that predates the risk-labeling scheme

Risk totals change with the catalog. Read the current labels from skills_index.json; they are declared metadata, not evidence that every procedure has been executed or audited.

Read SECURITY_GUARDRAILS.md and the security-guardrails contributor guide before invoking anything marked critical or offensive.

How to find the right skill

Search the full catalog — with AAS Core running, use search_skills/get_skill; see the usage guide.
Browse the generated catalog — CATALOG.md lists every skill by category with description, risk, source, and tags.
Use bundles as editorial shortlists — see bundles for curated packs like the Web Wizard.
Use the tool-specific guides — Claude Code, Codex CLI, Cursor, and Gemini CLI include starter picks.

Distribution surfaces

The same skills reach you through several channels, all backed by the same source tree:

Full library — direct install of the complete catalog (~/.agents/skills/ by default).
Plugins — root and bundle plugins for Claude Code and Codex (see plugins).
AAS Core — agent-owned, reproducible selection validated against the catalog.

Usage guide — compose and use a skill stack.
AAS Core — agent-owned catalog search and selection.
Discovery manifest — the machine-readable skills_index.json contract.
Bundles — curated skill packs.
Security skills — guidance for security-relevant skills.

This site is open source. Improve this page.