Awesome Claude Skills (Composio)
ComposioHQ/awesome-claude-skillsA curated list of Claude skills plus hundreds of templated app-automation skills that run through Composio's hosted Rube MCP.
With caveats
Broad access Depth 1/5
Best Agent Skills (Linkly) Claude skill
A one-skill pack plus daily-updated CSV rankings of the top 100 agent skills, merged from skills.sh, ClawHub, Tencent SkillHub and social mentions.
Our verdict Recommended with caveats
Useful open data with transparent methodology, but popularity is not quality and the skill depends on live fetches from the repo.
What Best Agent Skills (Linkly) does
Best Skills is mostly a dataset. The repo publishes nine top-100 rankings every day as CSV files with raw per-platform counts kept next to each score: an overall worth-installing score, top installs, 7-day trending, social buzz, most active, official skills, official vendors, top repos and rising stars. Data is collected from public endpoints of skills.sh, ClawHub and Tencent SkillHub plus GitHub and social sources, with history by date since mid-August 2026. The single SKILL.md (about 2 KB) lets an agent answer questions like which skill to install for PDFs or browser automation, what is trending, or whether a skill is popular or official. Its workflow is to fetch llms.txt (the column dictionary and rules) from the repo's main branch, pick the matching CSV, and answer with rank, raw numbers, vendor and install command. Rules say never to add install counts across platforms, to treat an empty cell as not measured, and to credit the data under CC BY 4.0. It recommends and links; it does not install anything.
Good for
- Asking an agent what is popular or trending in the skills ecosystem
- Comparing skills by installs across several registries
- Researchers who want open CSV data on skill popularity
Not for
- Judging skill quality or security (the numbers are popularity, buzz and maintenance signals)
- Offline use, since the skill fetches live files
What's inside depth 2/5
A 2.3 KB SKILL.md that points at data files; the repo's 1,400 files are mostly dated CSV snapshots rather than skill content.
- Skills
- 1
- Reference docs
- 11
- Scripts
- 0
- Tests / evals
- 0
- Templates
- 1
- Avg SKILL.md
- 2.3 KB
Counted from the repository file tree. How we score depth.
Dependencies and costs
- Network access to raw.githubusercontent.com to fetch llms.txt and the CSVs
Security review
Review before installingA data-lookup skill that answers 'which agent skill should I install' from daily ranking CSVs published in the author's GitHub repository. It has no scripts and no allowed-tools; the agent fetches llms.txt (a data dictionary) and CSV files from raw.githubusercontent.com at run time. The repository also contains maintainer GitHub Actions workflows that use secrets for publishing and alert emails, which do not run on your machine.
Worth knowing
- Makes network requestsDownloads ranking CSV files over HTTPS from raw.githubusercontent.com/LinklyAI/best-skills (no authentication, no data sent); the skill only recommends and links, it does not install anything.SKILL.md:22
What we checked
The repository (1422 files) on Oct 29, 2026. Not a full review: 1407 of 1422 files were not downloaded (large repository; we reviewed the skill folders); it relies on code outside the repository: 1407 of 1422 repository files were not available to us (the daily ranking CSV data); the daily-ranking pipeline runs from a separate LinklyAI/best-skills-runner repository that is not included.
NVIDIA SkillSpector static scan: 12 pattern hits (Data Exfiltration 6, MCP Rug Pull 4, analysis-evasion 1, Prompt Injection 1). The 12 SkillSpector hits are in maintainer files, not the skill: GitHub Actions workflows (daily.yml, heartbeat.yml) that send failure alert emails through the Resend API with repository secrets, npx skills mentions in docs, and an ordinary run of 83 spaces in a table row of the Russian README (no hidden characters found).
A review, not a certification. Skills change after we check them — read the files before installing.
License
CC BY 4.0 for data and documentation — commercial use allowed with attribution.
Can it run unattended?
Not really. A question-and-answer lookup skill with no loop or verification structure.
Alternatives to Best Agent Skills (Linkly)
skills (the npx skills CLI)
vercel-labs/skillsVercel Labs' open-source CLI (npx skills) for installing, updating and removing agent skills across Claude Code, Codex, Cursor, OpenCode and dozens more agents.
Recommended
Check first Depth 3/5
anbeime Skill Store
anbeime/skillA Chinese-language skill store: 63 local skills (content, video, e-commerce, PPT) plus daily-synced mirrors of other authors' skills.
Skip
Broad access Depth 2/5
FAQ
What is LinklyAI best-skills?
A daily-updated set of nine top-100 rankings of agent skills, published as CSV with raw install counts from skills.sh, ClawHub and Tencent SkillHub, plus social buzz, and a small skill that lets an agent query them.
How do I use best-skills with Claude Code?
Install the skill from the repo, or paste the one-line prompt from the README into your agent so it reads llms.txt and the CSVs. It answers with ranks and install commands but does not install anything itself.
Is best-skills safe?
It has no scripts. It fetches text and CSV files from the repository at run time, so you rely on the maintainers and on the agent treating fetched content as data.
Are the rankings a measure of quality?
No. They combine installs, growth, social mentions, maintenance and trust signals. Cross-platform counts are shown side by side and not added together, and a skill missing from a list is not necessarily low in installs.
Sources: GitHub repository (stars, license, file tree, last push Oct 29, 2026). Review based on the README and SKILL.md files as published.