ad-creative helps an agent write ad copy and pick what to run or kill. It defines angles, generates headline and description variations within platform character limits, and organizes the output for upload, including bulk CSV.
It also works from performance data: it analyzes winners and losers, generates new variations and logs each iteration. Further modes cover scaled static batches, creative strategy loops and a creative review page for client approval.
What You Get
- Headlines and descriptions grouped by angle, validated against platform character limits
- A bulk CSV layout ready for upload
- An iteration report with a performance summary, new creative and recommendations
- Test hypotheses that name the metric, comparison and expected direction
- An HTML creative review page template for stakeholder approval
Common Use Cases
- Writing responsive search ad headlines for a new campaign
- Refreshing Meta creative once the current ads fatigue
- Generating a static ad batch across several angles and sizes
- Preparing an approval page to share ad concepts with a client
What Makes It Unique
It sets writing standards that a test can lose: hypotheses must name the metric and direction, and copy must avoid AI tells such as the it’s-not-X-it’s-Y construction. Its references go deep on hook systems, awareness-level headlines, Meta formats, motion and short-form video, iMessage-style video ads and static templates.
Under The Hood
The SKILL.md loads first and pulls in reference files on platform specs, hooks, templates and generative tools, plus an HTML review page template. A tool section shows a command that runs a Node CLI from the repository’s tools folder to pull Google Ads performance data, and lists other ad platform CLIs and reporting connectors; those need your own ad account credentials. The skill itself ships no scripts.