Best AI Skills Tools 2026

OpenAI's public skills catalog for Codex-style skill packages, including reusable workflows such as document, spreadsheet, slide, image, and browser automation skills.

Best for: Codex users who want vendor-maintained examples before installing community skills.

Anthropic's official skill repository for Claude, with example skill folders and practical packages for document, spreadsheet, PDF, presentation, and creative workflows.

Best for: Claude Code teams that want skills aligned with Anthropic's documented format.

Current Codex documentation for skill structure, progressive disclosure, installation locations, marketplace skills, plugin skills, and authoring guidance.

Best for: Learning how Codex discovers, loads, installs, and shares skills.

Anthropic's Claude Code skill documentation for discovery, required frontmatter, installation folders, plugin marketplace distribution, and the skills command.

Best for: Validating whether a skill pack is structured correctly for Claude Code.

A higher-level Anthropic support overview of what Claude skills are, how they package instructions and files, and when users should apply them.

Best for: Explaining skills to non-technical stakeholders before rollout.

Codex's built-in skill installation flow, triggered with $skill-installer, for curated skills, GitHub repo skills, and marketplace/plugin skill sources.

Best for: Installing Codex-compatible skills without hand-copying folders.

Claude Code's in-product skills command for browsing, installing, and managing skills from supported skill locations and plugins.

Best for: Teams standardizing Claude Code skill installation.

A registry and open standard initiative for sharing agent skill packages across compatible AI coding tools.

Best for: Discovering cross-agent skill packages with a consistent vocabulary.

Community helper for creating Claude-compatible skill packages from guided prompts and packaging them for reuse.

Best for: Turning internal SOPs into repeatable skill folders.

Template repositories and generators that scaffold SKILL.md, references, scripts, and test fixtures for new custom skills.

Best for: Building private skills before publishing them to a team repo.

Community-maintained lists of Claude skill packages, example repositories, and specialized workflow packs.

Best for: Finding new skill ideas before choosing a vetted source.

GitHub topic page for agent skill repositories, including Claude, Codex, and cross-agent SKILL.md projects.

Best for: Broad discovery when you know the workflow category but not the package name.

Searchable registry experience for browsing skill packages by use case, compatible agent, and source repository.

Best for: Comparing installable skill packages across agent ecosystems.

Community directories that organize skills as plain Markdown packages, often optimized for easy copying into agent skill folders.

Best for: Lightweight private experiments and small team workflows.

Curated lists and newsletters tracking new Claude Code and Codex skill packages as the ecosystem changes.

Best for: Monitoring new skill packs without installing them immediately.

Skill packages for working with spreadsheets, slide decks, Word documents, PDFs, and other office files through agent workflows.

Best for: Finance, operations, reporting, and content production teams.

Skills that teach an agent how to use browser automation for navigation, screenshots, form testing, visual QA, and local app checks.

Best for: Frontend verification and repeatable product checks.

Skill packs that encode frontend implementation preferences, design-system rules, screenshot QA, and UI delivery checklists.

Best for: Product teams that want consistent agent-built interfaces.

Skills for analysis workflows that combine instructions with reusable scripts for charts, CSV cleanup, notebooks, spreadsheets, and reports.

Best for: Analysts who want reliable repeated transformations.

Custom skills for code review, bug reproduction, test writing, migration playbooks, release checks, and repository-specific engineering standards.

Best for: Teams turning senior engineering habits into repeatable agent behavior.

Skills that package editorial rules, SEO templates, content briefs, citation workflows, image generation steps, and publishing QA.

Best for: Content teams building consistent AI-assisted production pipelines.

Your own skill folders for internal SOPs, brand rules, deployment runbooks, compliance checks, and repository-specific practices.

Best for: Teams that need durable agent behavior tied to local context.

A practical pre-install review: read frontmatter, inspect scripts, check external commands, verify references, and install into the narrowest scope needed.

Best for: Avoiding risky third-party skill installs.

Use repository metadata, commit history, issue activity, license files, and script diffs to judge whether a public skill pack is safe enough to try.

Best for: Screening community skill repositories.

Keep sample inputs, expected outputs, and short acceptance checks beside custom skills so teams can verify behavior after edits.

Best for: Maintaining private skills over time.