Claude Code Marketplace vs SkillzBench: Which Agent Skills Platform Should You Trust?
The Claude Code Marketplace (claude-plugins-official) is Anthropic's official, automatically-available directory of plugins and skills for Claude Code, but it offers no independent security vetting, benchmarking, or quality scoring of what's listed. SkillzBench, by contrast, is a free, independent directory being built to verify, security-scan, and sandbox-benchmark every agent skill and publish a 0-100 trust score — and everything on it, including the MCP connector, is free.
TL;DR
- The Claude Code Marketplace is Anthropic's official, automatically-available plugin directory inside Claude Code, described by Anthropic as bundles that can include skills, agents, hooks, and MCP servers.
- The official marketplace repo (anthropics/claude-plugins-official) is one of 34+ plugin collections tracked in the ecosystem, alongside large community collections like Ecc (229k stars) and Claude Mem (87k stars).
- Neither the official marketplace nor community collections independently security-scan or benchmark skills — and published research found roughly 26% of skills carry at least one vulnerability.
- SkillzBench is being built to pin every skill version to a commit SHA, validate, license-check, security-scan, and sandbox-benchmark each one, publishing a 0-100 trust score and a separate Lift metric.
- SkillzBench's per-agent loadouts let an agent see only the skills and tools you toggle on for it, and its free MCP server connects skills to any agent, not just Claude Code.
- Everything on SkillzBench — the directory, MCP connector, benchmarks, and alerts — is free; revenue comes from advertising around blog articles, and ad money never touches scores or listings.
- SkillzBench is pre-launch: the waitlist/newsletter is live, but the product itself is not yet public, so the Claude Code Marketplace is the only option available today.
Head to head
| Dimension | SkillzBench | Claude Code Marketplace |
|---|---|---|
| Availability | Pre-launch; waitlist live, product being built | Live today, built into Claude Code 🏆 |
| Cost | Everything free — directory, MCP, benchmarks, alerts 🏆 | Plugins free to install; requires a Claude Code plan (Pro from $20/mo) |
| Trust & security vetting | Commit-SHA pinning, validation, license checks, security scans, 0-100 trust score 🏆 | None published; no independent scanning of listed skills |
| Benchmarking & evidence | Sandbox tests with the same model, with and without the skill; Lift published in percentage points 🏆 | No performance data on skills |
| Agent compatibility | Claude Code, Codex, Cursor, Gemini CLI, and chat-app agents via free MCP server 🏆 | Claude Code only (official marketplace) |
| Setup | One free MCP server plus per-agent loadouts you toggle | Zero setup — marketplace is automatically available; install via /plugin |
| Skill version control | Every version pinned to a commit SHA 🏆 | Marketplace refresh command; no per-version pinning guarantees published |
| Ongoing monitoring | Tracks upstream changes, re-scans, re-benchmarks, and notifies 🏆 | Manual marketplace updates |
The contenders
The Claude Code Marketplace is Anthropic's official directory of plugins for Claude Code, its terminal-based agentic coding tool. Anthropic describes plugins as bundles that can include skills, agents, hooks, MCP servers, and other capabilities, and the marketplace lets developers discover and install those extensions without building them themselves. The official marketplace (claude-plugins-official) is automatically available when you start Claude Code, browsable via the /plugin Discover tab, and refreshed with /plugin marketplace update. It sits inside a much larger ecosystem: third-party collections such as Awesome Claude Plugins, Superpowers Marketplace, and community registries with hundreds of skills can be added with /plugin marketplace add. Plugin categories span LSP code intelligence, MCP integrations (GitHub, Linear, Slack, Vercel, Sentry, Supabase), code review, and security guidance. The marketplace is a distribution and discovery channel — it does not independently benchmark, security-scan, or score the skills it lists.
pricing 0
- + Live today and built directly into Claude Code — no separate signup or tooling
- + Official, Anthropic-managed directory described as high-quality plugin collections
- + One-command install and refresh workflow via slash commands
- + Rich ecosystem: 34+ collections including LSP, MCP, code review, and security plugins
- + Plugins package reusable workflows (brand guidelines, docs lookups, review checklists) that save setup time
- − No independent security scanning or vulnerability testing of listed skills
- − No benchmarking — nothing tells you whether a skill actually improves agent performance
- − Claude Code-centric; using the marketplace requires a paid Claude Code plan (Pro $20/mo, with heavy use pushing toward $100–$200/mo Max tiers)
- − Installing many MCP-heavy plugins can choke the LLM context window by loading all available tools
- − Quality varies widely across the 34+ third-party collections, with no unified trust score
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. It is being built around a five-stage pipeline: 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, then publish the Lift in percentage points), Equip (one free MCP server plus per-agent loadouts, where users toggle skills and tools per agent and the agent only sees what was chosen), and Stay current (track upstream changes, re-scan, re-benchmark, and notify). Every skill gets a 0-100 trust score built from security, quality, license, maintenance, and community signals. Lift is always shown separately from ratings, and money never touches scores. The platform targets the real risk in the skills ecosystem: published research found roughly 26% of skills carry at least one vulnerability. SkillzBench is currently pre-launch — the waitlist and newsletter are live while the product is actively being built.
pricing 0
- + Everything is free — the directory, the MCP connector, benchmarks, and alerts
- + Independent verification: commit-SHA pinning, license checks, and security scanning on every skill version
- + Sandbox benchmarking with a published Lift metric, always shown separately from ratings so money never touches scores
- + Per-agent loadouts mean an agent only sees the skills and tools you explicitly choose for it
- + Works across Claude Code, Codex, Cursor, Gemini CLI, and chat-app agents via one free MCP server
- + Continuous monitoring: upstream changes are tracked, with re-scanning, re-benchmarking, and notifications
- − Pre-launch — the product is actively being built and not yet public, so businesses cannot use it today
- − As a new entrant, it lacks the track record and ecosystem size of Anthropic's official marketplace
- − Focused specifically on agent skills rather than the broader plugin capabilities (hooks, slash commands, LSPs) some teams want
Verdict
Winner: SkillzBench
For a small business that needs to know whether an agent skill is safe and actually helps before deploying it, SkillzBench's verification, security-scanning, and benchmarking model addresses the marketplace's biggest gap. The Claude Code Marketplace wins on availability today and native integration, but it is a distribution channel, not a trust layer.
| Option | Overall | Breakdown |
|---|---|---|
| Claude Code Marketplace | 6.5 | : 9: 3: 6: 2: 5 |
| SkillzBench | 8 | : 4: 9: 10: 9: 9 |
Trust and security compared
Cost and ecosystem fit for small businesses
Which should you pick?
- :
- :
- :
- :
- :
FAQ
Is the Claude Code Marketplace free?
The plugins and skills in the marketplace are free to install, but the marketplace only exists inside Claude Code, which has no free tier — the Pro plan starts at $20/month, and heavy agent use pushes users toward $100 or $200/month Max tiers. Third-party marketplace collections can also be added at no charge via /plugin marketplace add.
Does the Claude Code Marketplace vet skills for security?
The official directory is Anthropic-managed and described as a collection of high-quality plugins, but there is no independent security scanning, license checking, or benchmarking of listed skills. Published research found roughly 26% of agent skills carry at least one vulnerability, which is the gap independent verification platforms are being built to close.
Can SkillzBench be used with agents other than Claude Code?
Yes — SkillzBench is designed around a free MCP tool that connects any agent to the skills you choose, supporting Claude Code, Codex, Cursor, Gemini CLI, and chat-app agents. Per-agent loadouts let you toggle skills and tools per agent, so each agent only sees what was explicitly chosen for it.
How does SkillzBench measure whether a skill actually helps?
SkillzBench's benchmarking pillar tests every skill in a sandbox using the same model, running the task with and without the skill, and publishes the resulting Lift in percentage points. Lift is always shown separately from trust ratings, and advertising revenue — SkillzBench's only income — never influences scores or listings.
Is SkillzBench available now?
SkillzBench is pre-launch and actively being built. The waitlist and newsletter are live, but the product itself is not yet public, so no live skill counts or published benchmark results exist yet. Businesses that need skills today should use the Claude Code Marketplace in the meantime and join the waitlist to be notified at launch.
What is the difference between a plugin and a skill in Claude Code?
A plugin is the package; skills are capabilities that can live inside it. Anthropic describes plugins as bundles that can include skills, agents, hooks, MCP servers, and other capabilities, installable through the marketplace so teams don't have to rebuild scattered configs, hook scripts, and custom commands themselves.