An AI system that proposes App Store moves against live signals.
A weekly agent that reads App Store data, search trends, and your own analytics, and writes a one-page memo with three concrete plays. Tripled my ranked keywords; DR 0 → 20 in 8 weeks.
In one paragraph.
The agentic ASO toolkit is a small Node + Claude wiring that pulls App Store data, search-trend data, RevenueCat, and your own analytics, then writes a structured weekly memo: three proposed plays, evidence behind each, risk note, recommended action.
It shares a pattern with the Drops internal agent (CASE.01), an agent wired to live data sources with custom skills, but aimed at a single job: the weekly ASO memo. Lives on a cron. Costs about 40 cents a day to run.
The motivation.
I built it for StillMind because doing ASO properly is a part-time job I didn’t have time for. Once it was working, I open-sourced it because the wiring is the interesting part — not the prompts, which need to be reshaped per app anyway.
The features that matter.
Weekly memo engine
Reads data sources, writes a structured memo, opens a PR with the recommended changes.
Memo templates
Three plays + evidence + risk + recommended action. Strict format makes it usable.
Memory layer
Reads previous memos and logged outcomes; self-corrects across weeks.
MIT-licensed
Fork it, ship it, send PRs. No attribution required.