Use it when
- You need a practical growth, feedback, subject-line, or distribution workflow for a real newsletter job.
- You want the AI agent to turn audience context into a measurable plan, not generic marketing advice.
- You need clean distinctions between experiments, recommendations, and claims.
What you need before running it
Audience segment, current channel, goal, baseline, campaign window, and constraints.
Existing newsletter issue, feedback, or distribution asset.
What can be measured now and what still needs instrumentation.
What comes out
- A usable audience plan, analysis, or message set with local relevance and measurement built in.
- Clear experiment steps, owner actions, and stop conditions.
- Risks and unknowns called out before the plan scales.
What’s in the package
SKILL.mdexamples/worked-example.mdreferences/quality-checks.mdThe archive also includes a manifest with version, checksum, compatibility, category, and package metadata.
Quality gates
- No bought-list tactics or dark-pattern growth advice.
- Consent, source quality, and retention quality are explicit.
- Forecasts are replaced with pilots when baseline data is missing.
Compatibility checked
Codex, Claude
Compatibility is published only after package structure and ordinary read-path checks.