Paste a Claude Code skill and this free linter checks the YAML frontmatter, the name and description rules, body length, and the when-to-use structure — then tells you exactly what to fix. Nothing leaves your browser.
Paste your SKILL.md to validate it live, or load an example to see the linter in action.
100% client-side · nothing uploaded · works for any Claude Code / Agent Skill
A skill only fires when its metadata is exactly right. The validator enforces the hard requirements and flags the conventions that separate a good skill from a great one.
Confirms the file opens with a fenced --- block, the block closes, and every top-level line is parseable key: value YAML.
Required, kebab-case only (lowercase, numbers, hyphens), max 64 characters, and a reminder to match the skill's folder name.
Required, under 1024 characters, third-person, and starting with a clear “Use when…” trigger so Claude loads it at the right moment.
Flags bloated skills and nudges you to move long detail into linked reference files — the pattern Anthropic recommends.
Checks for explicit boundaries so the skill triggers on the right tasks and stays quiet on the wrong ones.
Warns on unrecognised frontmatter keys, missing H1 titles, and empty bodies before they cost you a broken skill.
A Claude Code Agent Skill is a folder with a SKILL.md at its root. The file has two parts: YAML frontmatter and a Markdown body. Here's the anatomy.
Lowercase letters, numbers and hyphens only — like web-security-audit. No spaces, capitals, or underscores. Maximum 64 characters, and it should match the name of the folder the SKILL.md lives in.
The single most important field: it is all Claude sees when deciding whether to load your skill. Write it in the third person, keep it under 1024 characters, and start with the trigger — "Use when…" — followed by what the skill does.
Keep SKILL.md under ~500 lines. If you need more, split reference material (schemas, long examples, checklists) into separate files in the skill folder and link to them from the body.
Explicit boundaries are what stop a skill from firing on the wrong task. Add a short “When to use” and “When NOT to use” section so the model has crisp guardrails.
name and a description) followed by instructions that tell Claude how to perform a task. Claude reads the description to decide when to load the skill.--- frontmatter block, a missing name or description, a name that isn't lowercase kebab-case or exceeds 64 characters, or a description over 1024 characters. This tool flags each of these as an error.This validator is free. The ClaudeThings kits give you a whole library of production-grade agents, skills, and slash commands — every one written to the exact bar this linter enforces.
Battle-tested Claude skills & workflows for builders.
Turn Claude into your growth team.
Or grab all 89 agents, 103 skills & 181 commands in the Complete Bundle →