Skip to main content
Category: GTM Intelligence · Version: 1.0.0 · Author: opendirectory · Compatibility: All targets (claude, codex, opencode, gemini, anti-gravity, openclaw, hermes)
Give this skill a competitor’s App Store or Google Play URL. It collects their public low-star reviews, detects where their store description’s claims break against real complaints, and outputs a one-session brief: ranked complaint clusters, a broken promise map, landing page headline variants, and ad copy you can use tomorrow. No API key. No paid tools. One URL in, one ready-to-use brief out.

​
Install

​
What It Does

  1. Collects up to 200 public reviews from App Store or Google Play — free, no API key, no login required. (Note: App Store apps may sometimes return 0 reviews due to Apple RSS API limitations; Google Play is the primary supported path.)
  2. Filters to 1–3 star reviews and scores each by severity (star rating) and recency — the most recent failures get the highest weight
  3. Clusters complaints into 4–6 named themes using the reviewers’ own language — no abstracted labels like “stability issues”
  4. Detects broken promises: cross-references the competitor’s own store description against the complaint clusters and flags where their marketing overclaims reality
  5. Outputs a copy-ready brief: complaint leaderboard with verbatim quotes, broken promise map, landing page H1 variants, ad copy directions, and anti-claim warnings

​
Prerequisites

  • Python 3.9+
  • One pip package (free, open source, no API key):
  • An AI agent: Claude Code, Gemini CLI, GitHub Copilot, or any compatible tool

​
Usage

Pass any of these prompts to your agent:
The product_context line (what your product does) is optional. If provided, ad copy directions will reference your specific product features. If absent, directions are phrased as positioning templates.

​
Output

One markdown file saved to docs/review-briefs/[app-slug]-[YYYY-MM-DD].md. Five sections, always present:

​
When NOT to Use

  • Need to map a whole market category? Use map-your-market instead
  • Need to find what features users want built next? Use gh-issue-to-demand-signal instead
  • Need to write outreach emails after reading the brief? Use outreach-sequence-builder next
  • Need to analyze your own app’s reviews? This skill is for competitor analysis only — the workflow is different
  • App has fewer than 10 low-star reviews? The skill will stop and tell you — that’s by design

​
Plays Well With

​
Cost per Run

​
Data Quality Notes

  • All quotes in the output are verbatim from collected reviews — no paraphrase, no editing
  • Broken promise detection is only as good as the competitor’s store description. If the description is sparse, Section 2 will note this rather than fabricating comparisons
  • Collection is capped at 200 reviews. For apps with very high review volumes, this is a sample — the most recent reviews by default
  • The scoring formula weights recent 1-star reviews most heavily: complaint_weight = (4 - rating) × recency_factor
  • See references/scoring.md for the full formula and tier thresholds