# SkillCheck

> Validate Agent Skills for Claude Code, Cursor, VS Code, Windsurf, and the AI coding ecosystem. Catch structural issues, anti-slop, security problems, and accessibility violations before your users do.

Agent Skills are markdown files (SKILL.md) that give AI coding agents procedural knowledge. They follow the open standard at agentskills.io. SkillCheck validates these files against 112+ checks across 22 categories.

## Pages

- [Home](https://getskillcheck.com/): Product overview, pricing, FAQ
- [Try Online](https://validate.getskillcheck.com/): Browser-based WASM validator (Free tier, no server, no data sent)
- [Methodology](https://getskillcheck.com/methodology): How checks are designed and what research informs them
- [Changelog](https://getskillcheck.com/changelog): Full version history from v1.0 to v3.18.0
- [Validated Skills](https://getskillcheck.com/reports/): Public skill reports with scores and grades
- [GitHub (Free)](https://github.com/olgasafonova/skillcheck-free): MIT-licensed skill file for Claude Code, Cursor, Claude Desktop

## Install

### Free (skill file, no binary)
```
git clone https://github.com/olgasafonova/skillcheck-free ~/.claude/skills/skill-check
```
Then say "skillcheck my skill" in Claude Code.

### Pro (Go binary + MCP server)
Purchase at https://getskillcheck.com/#pricing ($79 lifetime). Download binary from purchase email. Add to ~/.claude.json with license key. All Pro checks unlocked automatically.

## What It Checks

Free: Structure, body, naming, semantics (wisdom/platitude detection, description trigger style, railroading), quality patterns (gotchas, examples, error handling), design pattern classification, knowledge density (hollow content detection)

Pro: Anti-slop detection, WCAG accessibility, security scanning (PII, credentials, path traversal), token budget analysis, enterprise readiness, agent readiness (L0-L3 maturity), workflow checks, trigger collision detection, eval kit generation, knowledge density strengths

## Pricing

- Free: $0 forever, MIT open source, runs as a skill file
- Pro: $79 lifetime license, Go binary, MCP server, 112 checks
- Team: $49/month (coming soon), up to 10 users, custom rulesets

## Key Facts

- Independent project, not affiliated with Anthropic, OpenAI, or Google
- Scanned 1,613 public skills across 8 repositories (average score: 65/100)
- Works as a Claude Code skill (Free) or MCP tool server (Pro)
- No external API calls, no cloud dependency, runs locally
- Built by Olga Safonova

## Optional

- [llms-full.txt](https://getskillcheck.com/llms-full.txt): Comprehensive documentation for agents