user-layer

# UserLayer UserLayer turns App Store and Google Play reviews into structured research outputs: - pain points - user segments - market opportunities - follow-up answers grounded in cited reviews Use the bundled wrappers in `scripts/main.py`. Do not hand-roll raw HTTP requests when this skill is available. ## Use This Skill When - The user gives you an App Store or Google Play URL and wants review-backed insights. - The user wants to ask follow-up questions about a completed full analysis. ## Prerequisites - `API_KEY` must be set to a valid UserLayer API key. - `LAUNCHBASE_API_URL` is optional. Default: `https://lb-api.workflowhunt.com` ## Credentials and Host - Primary credential: `API_KEY` - Default API host: `https://lb-api.workflowhunt.com` - `API_KEY` should be scoped only for UserLayer API access and treated as a paid production credential. ## Available Wrappers - `analyze(app_url: str, max_reviews: int | None = None)` - `check_status(analysis_id: str)` - `query(pain_point_id: str, question: str)` ## Typical Flow 1. Call `analyze()` to start a full async run. 2. Poll with `check_status()` until the report is complete. 3. Read the returned `pain_points`. 4. Use a `pain_point.id` with `query()` for targeted follow-up questions. You do not need a separate preview step. `analyze()` is the main public entry point. ## Pricing Baseline - One `analyze()` run costs `$2.99` and includes retrieval plus full analysis of the latest `100` reviews. - If `max_reviews` is raised above `100`, extra reviews are billed as add-ons at `$0.01` per extra review. - `query()` is billed at `$0.01 / 1K tokens`. ## Key Response Shapes - `analyze()` starts an async job and returns a lightweight payload with: - `success` - `data.analysis_id` - `data.status` - `check_status()` returns the full completed report when ready, including: - `data.pain_points` - `data.user_segments` - `data.opportunities` - `sources` - `usage` - `query()` returns a follow-up answer for a specific pain point, including: - `data.answer` - `data.confidence` - `sources` - `usage` Treat `pain_points[].id` from a completed analysis as the required input for `query()`. ## Operating Rules - Default review count is `100` latest reviews. - `analyze()` includes retrieval and analysis of `100` latest reviews by default. - If `max_reviews` is raised above `100`, extra reviews are billed as add-ons. - `analyze()` is asynchronous and returns an `analysis_id`; poll with `check_status()`. - Use `query()` only after a completed full analysis returns a valid `pain_point_id`. - Treat `sources` and cited review evidence as the source of truth. - If `DATA_INDEX_NOT_FOUND` is returned, rerun a full analysis before querying again.