Help us improve
Share bugs, ideas, or general feedback.
From skillforge
Use when you need to audit a skill against design best practices, token efficiency, and known gotchas
npx claudepluginhub jugrajsingh/skillgarden --plugin skillforgeHow this skill is triggered — by the user, by Claude, or both
Slash command
/skillforge:auditingThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Evaluate a skill's SKILL.md and reference files against best practices for structure, modularization, and token efficiency.
Creates p5.js generative art with seeded randomness, noise fields, and interactive parameter exploration. Use for algorithmic art, flow fields, or particle systems.
Share bugs, ideas, or general feedback.
Evaluate a skill's SKILL.md and reference files against best practices for structure, modularization, and token efficiency.
If path provided, read it directly. Otherwise:
Glob: **/skills/*/SKILL.md
Ask user which skill to audit via AskUserQuestion.
Read the SKILL.md and list any references/ directory contents:
Glob: {skill_dir}/references/*.md
Count SKILL.md lines. Reference: every SKILL.md line costs tokens on every invocation.
| Lines | Rating | Action |
|---|---|---|
| < 200 | Good | No action needed |
| 200-400 | Watch | Consider if all content is always needed |
| 400-500 | Extract | Identify conditionally-loaded sections |
| > 500 | Over budget | Must extract to references |
Read references/structure.md for the structural checklist.
Read references/modularization.md for the reference file pattern checklist.
Read references/gotchas.md for known pitfalls.
============================================================================
Skill Audit: {plugin}:{skill}
============================================================================
Token Budget:
SKILL.md: {lines} lines ({rating})
References: {count} files, {total_lines} lines (loaded conditionally)
Structure: Score
✓/✗ Frontmatter complete {pass/fail}
✓/✗ Skill has allowed-tools {pass/fail}
✓/✗ Workflow numbered steps {pass/fail}
✓/✗ Detection before generation {pass/fail}
✓/✗ User confirmation before action {pass/fail}
✓/✗ Report section at end {pass/fail}
Modularization: Score
✓/✗ No unconditional large blocks {pass/fail}
✓/✗ References used for variant content {pass/fail}
✓/✗ References loaded conditionally {pass/fail}
✓/✗ No repeated content across references {pass/fail}
Gotchas: Score
✓/✗ No backticks around shell metacharacters {pass/fail}
✓/✗ No name collision (command vs skill) {pass/fail}
✓/✗ Variables use correct placeholder syntax {pass/fail}
✓/✗ ASCII indicators (no emoji) {pass/fail}
Findings:
{severity} {description}
...
Recommendations:
{numbered_list}
============================================================================
references/structure.md - Skill structure checklistreferences/modularization.md - Reference file patterns and when to extractreferences/gotchas.md - Known pitfalls in skill authoring