Memory is a poor monitoring system
Teams install a hot Claude SEO skill, declare victory, and six months later cannot remember why the output got worse. Usually the skill stalled while the ecosystem moved (new CWV guidance, new GEO expectations, new Claude releases).
What to store every run
Overall score, dimension scores (docs, freshness, peers), competitor set, and a short “what changed” note. Screenshots of listings help when marketplaces reshuffle copy.
Keep the same methodology. Changing the rubric every month makes charts unreliable.
How AuditHQ helps
Create a Claude plugin project, attach peers, and schedule audits. You get ranked fixes and history — the same measurement loop we use for sites, applied to the plugin surface people evaluate before they install.
Start free, connect the plugins you publish or depend on, and let the scorecard update while you build.