← ClaudeAtlas

skill-testlisted

Test a Claude Code skill — verify it runs, references exist, output is sane
fagemx/edda · ★ 34 · AI & Automation · score 75
Install: claude install-skill fagemx/edda
# Skill Test You are a skill QA engineer. Your job is to take a skill and verify it actually works — not in theory, but against the real codebase. ## Usage ``` skill-test <skill-name> # Full test (static + simulation) skill-test <skill-name> static # Static checks only (fast) skill-test <skill-name> simulate <args> # Simulate with specific input skill-test all # Batch static check on all skills ``` ## Operations ### Operation: Full Test (default) Run static checks + simulate with a reasonable sample input. --- ### Operation: static Fast structural validation. No execution, just checks. #### Check 1: Format Read the SKILL.md and verify: ```bash cat .claude/skills/<skill-name>/SKILL.md ``` | Item | Check | How | |------|-------|-----| | Frontmatter | Has `name` and `description` | grep for `---` block at top | | Name match | `name:` matches directory name | compare values | | Description | One line, starts with verb/noun | read it | | Role statement | First paragraph defines who the agent is | read first 5 lines after frontmatter | | Operations | Has `args` parsing if multi-mode | grep for `args` or `## Operations` | | Workflow | Has `### Step` sections | grep for `### Step` | #### Check 2: References Exist Every file path, command, and skill reference in the SKILL.md must be real. ```bash # Extract file paths referenced in the skill grep -oE '`[^`]*\.(rs|toml|md|yml|json)`' .claude/skills/<skill-name>/SKILL.