Primary question: Are the official-team skills here (Stripe, Supabase, HashiCorp Terraform, Expo) worth adding to your coding agent before you write your own?
RepoDaily adoption score
RepoDaily rates this as 91/100 (strong) for adoption: evidence, installation path, production risk, differentiation, license clarity, and AI/agent fit are scored from the article sources and adoption notes.
4 source(s) across 3 source category/categories, plus a RepoDaily-specific evidence module when available.
4 workflow step(s), 5 next-action step(s), and 1 command/install signal(s) were detected.
Trending momentum is +600 stars, with maintenance/release/issue signals counted when present.
Risk is marked low, with 5 security note(s) and 3 explicit skip condition(s).
3 opportunity lens item(s), 4 alternative(s), and 4 type-specific section(s) support differentiation.
License source or license wording is present.
8 AI/agent-related signal(s) were detected in the article text and metadata.
Project overview
VoltAgent/awesome-agent-skills is an awesome-style index, not a package: CONTRIBUTING.md states plainly that 'This repository curates links only. Each skill lives in its own repo.' What you get is a single README organized by publisher. The repo description advertises '1000+ agent skills,' while the README badge already counts 1497+, and the maintainers draw a hard line against the flood of generated lists - the page is tagged 'Hand-picked, not AI-slop generated.' Official sections come from Anthropic, Google Labs, Vercel, Stripe, Cloudflare, Netlify, Trail of Bits, Sentry, Expo, Hugging Face, and Figma, sitting alongside community-submitted skills.
The pitch is cross-tool coverage. The README names eight compatible coding agents - Claude Code, Codex, Antigravity, Gemini CLI, Cursor, GitHub Copilot, OpenCode, and Windsurf - and points readers to its table 'for paths and documentation' rather than shipping any installer. The table of contents reads like a vendor directory: Official Claude Skills, then sections for VoltAgent, TestMu AI, Modem Dev, Angular, Composio, Supabase, Google Gemini, Stripe, Courier, CallStack, Expo, Better Auth, Tinybird, HashiCorp (for Terraform), Sanity, Firecrawl, Neon, ClickHouse, Remotion, and Replicate. Community entries are bucketed into Marketing, Productivity and Collaboration, Development and Testing, Context Engineering, AI and Data, n8n Automation, and Other.
Curation is rule-based rather than vibe-based. Every entry must point to a public repository with a working skill, carry documentation (a README or SKILL.md), include an author/org prefix in its name, and describe itself in ten words or fewer. The gate that matters most: a skill 'must have real community usage,' and 'brand new skills that were just created are not accepted.' Submissions append one markdown line to the end of the relevant section and open a PR titled 'Add skill: author/skill-name.' Reviewers state they 'may decline skills that don't meet the quality bar.'
Traction and governance are both visible. The repo picked up 600 stars in the 2026-08-25 trend window at rank 9. The list is MIT licensed (Copyright (c) 2025 VoltAgent), maintained under the VoltAgent organization, with a last-commit badge and a Discord invite in the README. Sponsorship from TestMu AI (formerly LambdaTest) and Modem is displayed in a dedicated section, separate from the skill sections; the documented contribution rules remain the published bar for inclusion.
Why it is trending now
- 600 stars in the trend window and rank 9 on the 2026-08-25 list.
- The README badge counts 1497+ skills, ahead of the '1000+' in the repo description - a concrete sign the index keeps growing.
- Official vendor participation: Anthropic, Google Labs, Vercel, Stripe, Cloudflare, Netlify, Trail of Bits, Sentry, Expo, Hugging Face, and Figma all have named sections.
- One index covers eight named coding agents (Claude Code, Codex, Antigravity, Gemini CLI, Cursor, GitHub Copilot, OpenCode, Windsurf), removing per-tool hunting.
- Anti-slop positioning backed by rules: brand-new skills are rejected and real community usage is required before an entry is accepted.
Problem it solves
- Skill discovery is noisy: the README explicitly calls out 'bulk-generated skill repositories' and 'mass AI-generated stuff' as the problem it replaces.
- Vendor skills are scattered across separate official repos - Stripe, Supabase, HashiCorp, and Expo each publish their own - with no shared index of tool compatibility.
- Trust signals are thin elsewhere: entries here must carry documentation (README or SKILL.md), an author/org prefix, and evidence of community adoption.
How it works
- Open README.md and jump to a vendor section (e.g., 'Official Claude Skills' or 'Skills by Supabase Team').
- Each entry is one markdown line in the form [author/skill-name](https://github.com/author/repo/path) with a description of ten words or fewer.
- Click through to the skill's own repository, read its README or SKILL.md, and follow that repo's instructions for your tool (Claude Code, Codex, Gemini CLI, Cursor, and so on).
- To contribute, append your entry at the end of the matching category and open a PR titled 'Add skill: author/skill-name'.
Product demo and interface preview
What the 1,497+ entries actually are
Each entry is a single markdown line: a bolded link shaped like [author/skill-name](https://github.com/author/repo/path) followed by a description capped at ten words. There is no registry, no package integration, and no code in this repository - just links to repos that each carry their own license and documentation. That makes the list cheap to use as a directory and impossible to treat as a single vendor: quality control happens at review time, not at install time.
Integration surface: eight tools, one index
- Named compatible agents in the README: Claude Code, Codex, Antigravity, Gemini CLI, Cursor, GitHub Copilot, OpenCode, Windsurf - plus 'and more'.
- The README defers specifics to its own table 'for paths and documentation'; no install command lives in this repo.
- Vendor sections currently in the table of contents: Claude, VoltAgent, TestMu AI, Modem Dev, Angular, Composio, Supabase, Google Gemini, Stripe, Courier, CallStack, Expo, Better Auth, Tinybird, HashiCorp (Terraform), Sanity, Firecrawl, Neon, ClickHouse, Remotion, Replicate.
- Community subcategories: Marketing; Productivity and Collaboration; Development and Testing; Context Engineering; AI and Data; n8n Automation; Other.
- Every install path routes through the linked skill's own repository and its README or SKILL.md.
Maintenance read: the rules that keep the list usable
- CONTRIBUTING.md requires each entry to point at a public repository with a working skill and documentation (README or SKILL.md).
- Descriptions are capped at 10 words and must include an author/org prefix - both enforced at review.
- Brand-new skills are rejected outright: 'Give your skill time to mature and gain users before submitting.'
- Reviewers 'may decline skills that don't meet the quality bar,' so acceptance is a judgment call, not automatic.
- The README carries a Skills badge (1497+) and a GitHub last-commit badge, so growth and freshness are visible at a glance.
- Because the repo is links-only, dead links are the main structural risk; CONTRIBUTING asks submitters to 'Verify your links work before submitting.'
Adoption checklist before installing any linked skill
- Confirm your agent is in the compatibility list (Claude Code, Codex, Antigravity, Gemini CLI, Cursor, GitHub Copilot, OpenCode, Windsurf).
- Open the skill's own repo and read its README or SKILL.md end to end - both are required by CONTRIBUTING, so a missing doc is a disqualifier.
- Check that repo's license separately: the MIT license here (Copyright (c) 2025 VoltAgent) covers the list, not the linked skills.
- Look for the same evidence maintainers demand - real community usage, not a repo created last week.
- Prefer vendor sections (Stripe, Supabase, HashiCorp Terraform, Expo, Cloudflare, Netlify) when you want first-party documentation behind the skill.
Alternative matrix: how this index differs
- anthropics/skills: single-vendor and Anthropic-authored; the right pick if you only run Claude Code and want first-party skills.
- hesreallyhim/awesome-claude-code: Claude Code-centric and broader in scope (commands, MCP servers), but narrower in tool coverage than this repo's eight named agents.
- obra/superpowers: an installable bundle of skills rather than a link index - heavier commitment, less picking and choosing per skill.
- Cursor Directory: a dedicated website for Cursor rules; useful for Cursor users, but not a cross-tool index with vendor sections.
Who should pay attention?
Good fit if
- You run Claude Code, Codex, Gemini CLI, Cursor, GitHub Copilot, OpenCode, Windsurf, or Antigravity and want vendor-authored skills (Stripe, Supabase, HashiCorp Terraform, Expo, Cloudflare, Netlify) in one place.
- You maintain a mature community skill with existing users and want discoverability via a PR titled 'Add skill: author/skill-name'.
- You are designing an internal approved-skills list and want a copyable structure - MIT licensed, so reuse is fine with the copyright notice kept.
Skip for now if
- You want an installable package or CLI - this repository ships links, not code.
- You just published a brand-new skill: CONTRIBUTING rejects entries without real community usage.
- You need uniform licensing - each linked skill carries its own license in its own repo.
Risks and cautions
The index itself is low-risk: MIT-licensed, links-only, with no executable content. Risk transfers to each third-party skill you choose to install from it.
- LICENSE is MIT (Copyright (c) 2025 VoltAgent) and covers the list, not the linked skills.
- CONTRIBUTING states the repository curates links only; every skill lives in its own repo with its own maintainers.
- Reviewers screen submissions and can decline entries, but they do not audit the code inside every linked repo.
- At 1497+ entries, per-link verification falls to the reader; spot-check each skill's README or SKILL.md before installing.
- This repository contains markdown links, not executable skill code; the list itself is MIT licensed (Copyright (c) 2025 VoltAgent).
- Every entry resolves to a third-party repository that you install into an agent such as Claude Code, Codex, or Cursor - read the skill's README or SKILL.md before installing it.
- CONTRIBUTING requires documentation (README or SKILL.md) for every entry, which gives you a review artifact before any install.
- Link verification is the submitter's responsibility ('Verify your links work before submitting'), so stale or moved links are possible.
- Sponsor logos (TestMu AI, Modem) appear in a dedicated README section; sponsorship is displayed separately from the skill listings.
Alternatives to compare
| Approach | When to use | Trade-off |
|---|---|---|
anthropics/skills | You only run Claude Code and want Anthropic-authored, first-party skills. | Free, open source |
hesreallyhim/awesome-claude-code | You want a Claude Code-focused list that also covers commands and MCP servers. | Free, open source |
| You prefer one installable skills bundle over browsing a 1,497-entry index. | Free, open source | |
Cursor Directory | You want Cursor-specific rules and skills from a dedicated site. | Free website |
What this trend reveals
Vendor-authored skills as a documentation pattern
The official sections - Stripe, Supabase, HashiCorp (Terraform), Expo, Cloudflare, Netlify - show how product teams write SKILL.md files for their own platforms. Reading three of them beats inventing a format from scratch.
Pick three vendor entries, open their README or SKILL.md, and compare section order, command naming, and guardrails.
A copyable skeleton for an internal approved-skills list
The list is MIT licensed (Copyright (c) 2025 VoltAgent), so you can reuse its structure - vendor sections plus community subcategories like Development and Testing or Context Engineering - for an internal shortlist, keeping the copyright notice as MIT requires.
Fork the structure, replace entries with internally approved skills, and confirm the MIT notice survives in your copy.
A publication channel for mature community skills
The contribution gate - real community usage, documentation, author/org prefix, ten-word description - means the list rewards skills that already have users, not launches.
Ship your skill, gather users, then append one line to the matching category and open a PR titled 'Add skill: author/skill-name'.
RepoDaily verdict
A genuinely useful index, not a product: 1,497+ linked skills, real vendor participation from Anthropic to Stripe, and contribution rules that filter out AI-slop. Use it as a discovery layer, then vet each linked skill on its own merits - the list's MIT license covers the links, not what they point to.