Group: GitHub Rongo270

Sort by:Popular▼

✓Popular✓Newest

One composable renders items as grid, list, cards or carousel - a per-user Thompson-sampling bandit learns which layout wins.

Fuses motion, screen, audio routes, DND, network and time into the user's current moment: driving, walking, working, in a meeting, just woke up.

Feature-discovery nudge engine: right feature, right moment, hard anti-nag guarantees and a dead-feature report.

Estimates why the user left: call, rage quit, task completed, bored, battery, screen off or crash - with evidence per verdict.

Feature usage tracking with a full built-in stats screen.

On-device UX analysis: mines orderings, drop-offs, confusion loops, funnel leaks and common paths, then emits recommendations.

Per-field form friction: corrections, dwell, refocus and where users give up, with concrete fix suggestions.

Detects handedness and thumb reach, heatmaps taps and flags controls users strain to reach, with a Compose debug overlay.

Understands why users act: rage taps, double-back, type-delete loops, scan-scrolling, failed drags, hesitation and a live frustration score.

User-felt performance: cold start, per-screen TTI, jank, tap latency and main-thread stalls with guilty stacks - one felt-score per screen.

Learns the user's weekly rhythm: best time to notify, next expected open, expected session length and churn risk.

Detects screenshots and classifies them on-device (receipt, error, chat, ticket, map, product, code), extracts entities and suggests actions.

Local user memory store: remembers facts about the user on-device.