feat: improve skill scores for claude-ads

[popey]

Hullo @AgriciDaniel 👋

I ran your skills through tessl skill review at work and found some targeted improvements. Here's the full before/after:

Skill	Before	After	Change
ads-landing	79%	100%	+21%
ads-competitor	75%	89%	+14%
ads-youtube	86%	100%	+14%
ads-creative	83%	93%	+10%
ads-audit	86%	93%	+7%
ads-google	86%	93%	+7%
ads-budget	86%	89%	+3%

Summary

Improved 7 skills with an average improvement of +7% (average score 85% → 92%). All 18 skills also had their frontmatter description format standardized from YAML block scalars (>) to quoted strings for consistency.

Type of Change

• Other (describe): Skill quality improvements — workflow clarity, actionability, and frontmatter standardization

Checklist

• Tested against a sample ad account (if applicable)
• SKILL.md files are under 500 lines
• Reference files are under 200 lines
• Python scripts output valid JSON
• Shell scripts use set -euo pipefail
• No credentials or API keys in the diff

Changes made

All 17 sub-skills

• Standardized frontmatter description from YAML block scalar (>) to quoted string format

ads-landing (+21%)

• Added validation checkpoints (URL accessibility, tracking pixel verification)
• Added executable speed test commands (Lighthouse CLI, PageSpeed API)
• Added conversion tracking verification code snippets
• Removed truisms, converted mobile checklist to checkbox format

ads-competitor (+14%)

• Added validation checkpoints (confirm competitors found, verify gap analysis uses real data)
• Removed redundant platform-specific research section
• Condensed competitive response strategies into compact playbook

ads-youtube (+14%)

• Added validation checkpoints (confirm active video campaigns, verify campaign types)

ads-creative (+10%)

• Added validation checkpoints (confirm platform data, verify fatigue signals)

ads-audit (+7%)

• Added validation checkpoints (confirm platform data, verify subagent scores)

ads-google (+7%)

• Added validation checkpoints (confirm 30-day data, verify all 74 checks evaluated)

ads-budget (+3%)

• Added validation checkpoints (confirm 14-day spend data, verify kill list data sufficiency)

ads (orchestrator)

• Added concrete subagent spawning example and validation step

ads-apple

• Added data validation checkpoint (confirm 30-day campaign data)

Honest disclosure — I work at @tesslio where we build tooling around skills like these. Not a pitch - just saw room for improvement and wanted to contribute.

Want to self-improve your skills? Just point your agent (Claude Code, Codex, etc.) at this Tessl guide and ask it to optimize your skill. Ping me - @popey - if you hit any snags.

Thanks in advance 🙏

[AgriciDaniel]

Thanks for this thorough work, @popey! The skill score analysis from Tessl was genuinely helpful.

After detailed review, I've cherry-picked the validation gate improvements from 6 skills:

• ads/SKILL.md — Task tool orchestration clarity + subagent JSON validation
• ads-audit — Platform data availability check + subagent score verification
• ads-budget — 14-day minimum for kill/scale + 20-click/$100 data sufficiency
• ads-creative — Data existence check + assumption prevention gate
• ads-google — 30-day data minimum + 74-check completeness verification
• ads-youtube — Active campaign check + campaign type completeness

Skipped: The YAML frontmatter reformatting (single-line descriptions) and the content compression in ads-competitor and ads-landing. The compression removed valuable platform-specific research instructions (Google Ads Transparency Center, Meta Ad Library steps), Lighthouse CLI commands, and trust element details that users rely on.

These will ship in v1.3.0. Appreciate the contribution and the transparent disclosure about Tessl! 🙌