Files

1.6 KiB

Experiment: Skill Description SEO Optimization

Objective

Optimize the description field in each skill's SKILL.md frontmatter to maximize:

  1. SEO discoverability — include high-volume keywords: "agent skill", "plugin", "Claude Code", "Codex", "Gemini CLI", "Cursor"
  2. Trigger accuracy — the description must accurately describe when the skill activates
  3. Clarity — one read should convey what the skill does and who it's for
  4. Cross-platform appeal — mention multi-tool compatibility where natural

Constraints

  • Description must be under 200 characters (plugin.json limit)
  • Must NOT be spammy or keyword-stuffed — natural language only
  • Must preserve the skill's actual purpose and capabilities
  • Do NOT modify anything outside the description: field in frontmatter
  • One skill per experiment iteration

Strategy

  • Start with the top 10 most-viewed skills (from GitHub traffic)
  • For each: read current description → rewrite with SEO keywords → evaluate
  • Terms to naturally incorporate: "agent skill", "plugin", "coding agent", tool names
  • Avoid: generic filler, "AI-powered", "comprehensive solution"

Target Skills (in priority order)

  1. marketing-skill/SKILL.md (all sub-skills)
  2. engineering-team/SKILL.md (all sub-skills)
  3. engineering/SKILL.md (all sub-skills)
  4. c-level-advisor/SKILL.md (all sub-skills)
  5. finance/SKILL.md (all sub-skills)
  6. business-growth/SKILL.md (all sub-skills)
  7. product-team/SKILL.md (all sub-skills)
  8. project-management/SKILL.md (all sub-skills)

Evaluation

Use llm_judge_content evaluator customized for SEO scoring. Metric: seo_quality_score (0-100) Direction: higher is better