Agenticskills.io Review 2026: How It Handles Skill Security and Licensing
Agenticskills.io is a free, editorially curated directory of 193+ AI agent skills and 196 MCP servers across 9 platforms, with a strong security angle — it publishes a 12-step MCP audit framework based on an analysis of 8 independent audits covering 23,000+ public MCP servers, and states that listings are never paid for. For small businesses that also want every skill version-pinned, license-checked, security-scanned and benchmarked for measurable Lift, SkillzBench is building a free directory and MCP connector that does exactly that.
TL;DR
- Agenticskills.io indexes 193+ curated agent skills, 196 MCP servers, and tracks 9 platforms (Claude Code, Codex, Cursor, Gemini CLI and more), with the index live at version 1.7.0 as of October 13, 2026.
- The site is free to use, states 'Listing is never paid for,' and explicitly labels directory judgments as 'editorial judgment, not a measurement.'
- Its security credentials are content-driven: an analysis of 8 independent audits covering 23,000+ public MCP servers found 73% fail basic security checks, paired with a 12-step audit framework readers can apply themselves.
- Skills are ranked with an S-rank metric (e.g., Anthropic's Frontend Design skill at 180.7K), and 100% of listed skills are marked open source.
- The learning library contains 11 guides, 8 topics, 177 minutes and 41K+ words of material, including a CLI guide verified against skills 1.7.0.
- The site does not state that it security-scans or license-checks each individual skill — its licensing signal is the 'Open Source' tag on listings.
- SkillzBench, currently in pre-launch, is building the missing piece: pinning every skill version to a commit SHA, license-checking, security-scanning, and benchmarking each skill's Lift in a sandbox.
Head to head
| Dimension | SkillzBench | Agenticskills.io |
|---|---|---|
| Pricing | Free — directory, MCP, benchmarks, alerts | Free; listings never paid for |
| Coverage | Continuously crawling and deduping skills from GitHub (pre-launch) | 193+ skills, 196 MCP servers, 9 platforms 🏆 |
| Security approach | Systematic: security-scan every skill, pin to commit SHA, re-scan on upstream changes 🏆 | Editorial: 12-step audit framework, 73% failure analysis of 23,000+ MCP servers |
| Licensing | License check on every skill as part of verification 🏆 | 'Open Source' tag on listings; license field on tool pages |
| Ranking method | 0-100 trust score from security, quality, license, maintenance and community signals 🏆 | S-rank popularity metric plus editor's picks |
| Benchmarking | Sandbox testing with and without each skill, publishing Lift in percentage points 🏆 | None — site states listings are 'editorial judgment, not a measurement' |
| Setup | One free MCP server connects any agent to chosen skills; per-agent loadouts | Browse freely; install skills via npx skills CLI (guide verified against skills 1.7.0) |
| Availability | Pre-launch; waitlist live | Live now (index v1.7.0, Oct 13 2026) 🏆 |
The contenders
Agenticskills.io is a curated directory of AI agent skills and MCP servers, live at index version 1.7.0 as of October 13, 2026. It indexes 193+ agent skills across 16 categories and 9 platforms — Claude Code, OpenAI Codex, Cursor, Gemini CLI, Claude AI, Windsurf and others — alongside 196 MCP servers, including sandbox platforms like Fly.io, E2B, Modal and Daytona. Every listed skill is tagged open source, and entries carry an S-rank popularity figure (Anthropic's Frontend Design skill, for example, shows 180.7K). The site is transparent about methodology: it states listings reflect 'editorial judgment, not a measurement' and that 'listing is never paid for.' Its editorial output leans heavily into security, most notably an analysis of 8 independent audits covering 23,000+ public MCP servers that found 73% fail basic security checks, plus a 12-step framework for auditing any server before installation. A learning library offers 11 guides, 8 topics, 177 minutes and 41K+ words of tutorials, including a guide to the npx skills CLI verified against skills 1.7.0.
pricing 0
- + Free to browse, with an explicit 'listing is never paid for' policy
- + Broad coverage: 193+ skills, 196 MCP servers, 9 platforms tracked
- + Serious security editorial: 73% failure-rate analysis across 23,000+ MCP servers and a practical 12-step audit framework
- + Transparent methodology — openly states rankings are editorial judgment, not measurement
- + Deep, current learning library (11 guides, 41K+ words) including CLI documentation verified against skills 1.7.0
- + Covers the emerging Agent Plugins open standard agreed by OpenAI, Microsoft, AWS, GitHub, Cursor and Vercel
- − No stated per-skill security scanning or license validation — the only licensing signal is an 'Open Source' tag
- − S-rank is a popularity metric, not a measured performance or safety score
- − Pricing for listed third-party tools is shown only as a model ('usage-based'), not a rate, so budgeting requires vendor pages
- − No benchmarking of whether a skill actually improves agent output
SkillzBench is AIQ Labs' free, independent directory of AI agent skills plus a free MCP tool that connects any agent to the skills you choose — currently in pre-launch with a live waitlist. It is built around published research finding that roughly 26% of skills carry at least one vulnerability, and its pipeline is designed to close that gap: Discover (continuously crawl and dedupe skills from GitHub), Verify (pin every version to a commit SHA, validate, license-check, security-scan, and quality-judge), Benchmark (test every skill in a sandbox with the same model, with and without the skill, and publish the Lift in percentage points), Equip (one free MCP server plus per-agent loadouts where the agent only sees what was chosen), and Stay current (track upstream changes, re-scan, re-benchmark, notify). A 0-100 trust score is computed from security, quality, license, maintenance and community signals, and Lift is always shown separately from ratings. Everything on the platform — directory, MCP connector, benchmarks and alerts — is free; revenue comes from advertising around blog articles, and ad money never influences scores or listings.
pricing 0
- + Everything is free — directory, MCP connector, benchmarks and alerts
- + Security-first design: commit-SHA pinning, license checks and security scans address the ~26% vulnerability rate found in research
- + Measured Lift benchmarking rather than popularity ranking; money never touches scores
- + Per-agent loadouts mean an agent only sees the skills and tools you toggle on
- − Pre-launch — the product itself is not yet public, so features cannot be used today
- − No live skill counts or published benchmark results available yet
Verdict
Winner: Agenticskills.io
Agenticskills.io is a live, free resource with a genuinely security-conscious editorial stance and a deep learning library. Its main limitation is that it ranks skills by an S-rank popularity metric rather than by measured performance or security scans of each skill.
| Option | Overall | Breakdown |
|---|---|---|
| Agenticskills.io | 7.5 | : 10: 8: 7: 7: 6: 8 |
| SkillzBench | 7 | : 10: 7: 7: 9: 6: 7 |
How Agenticskills.io handles skill security
Licensing: what each approach tells you
Who Agenticskills.io is for — and where it falls short
Which should you pick?
- :
- :
- :
- :
- :
FAQ
Is Agenticskills.io free to use?
Yes. The directory, guides and newsletter are free — the newsletter is described as 'free forever, unsubscribe anytime, no spam,' and the site states that listing placement is never paid for. There is no indication anywhere on the site of a paid tier or user fee.
Does Agenticskills.io security-scan the skills it lists?
The site does not claim to run automated security scans on individual skills. Its security work is editorial — notably an analysis of 8 independent audits covering 23,000+ public MCP servers (73% failed basic checks) and a 12-step framework readers can use to audit servers themselves. For systematic per-skill scanning, commit-SHA pinning and license checks, SkillzBench is building that verification pipeline and is currently in pre-launch.
How does Agenticskills.io check licenses?
Skill listings carry an 'Open Source' tag — the directory reports 100% of its curated skills are open source — and MCP server tool pages display a structured license field, such as 'Proprietary' for Fly.io. However, the site does not describe per-version license validation. SkillzBench's model makes license-checking a verification step, attached to a specific commit SHA.
What is S-rank on Agenticskills.io?
S-rank is the popularity figure shown on each skill listing — for example, Anthropic's Frontend Design skill shows 180.7K and obra's Brainstorming shows 298.0K. It reflects adoption, not safety or measured performance; the site itself notes that listings are 'editorial judgment, not a measurement.'
How many guides does Agenticskills.io's learning library contain?
The learning library contains 11 guides across 8 topics, totaling 177 minutes and 41K+ words. Topics range from the npx skills CLI (verified against skills 1.7.0) to the Agent Plugins open standard and security audits of public MCP servers.
Which platforms does Agenticskills.io cover?
The directory tracks 9 platforms, including Claude Code, OpenAI Codex, Cursor, Gemini CLI, Claude AI and Windsurf. Skills are tagged with the platforms they support, and the index was live at version 1.7.0 as of October 13, 2026.
What is the alternative if I want measured proof a skill helps?
SkillzBench, currently in pre-launch, is building sandbox benchmarking that runs every skill with the same model, with and without the skill, and publishes the Lift in percentage points. Lift is always shown separately from its 0-100 trust score, and everything — directory, MCP connector, benchmarks and alerts — is free.