Mobile observability
Skill Amey-Thakur/AI-SKILLS/skills/mobile/mobile-observability
Plug-and-play skills and prompts for every AI coding agent
npx -y skills add Amey-Thakur/AI-SKILLS --skill mobile-observabilityAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 19 days oldThe repository was created 19 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Instrument mobile apps for crash-free rate, ANR and hang detection, and release health you can gate rollouts on. Use when setting up mobile monitoring or deciding whether a release is healthy enough to expand.
SKILL.md
2.5 KB, as published. Nobody here has run it
Mobile observability
You cannot attach a debugger to a customer's phone. Observability is the replacement: every question you will ask during an incident must be answered by data the app already ships home.
Method
- Anchor on crash-free sessions. Crash-free sessions (not users) is the rollout gate metric; set a floor (99.5% is a common bar) and halt expansion below it. Track it per release version, since blends hide a bad build behind a good installed base.
- Symbolicate everything, automatically. Upload dSYMs / mapping files in CI for every build, including bitcode-stripped and R8-minified ones. An unsymbolicated stack trace is a support ticket, not a diagnosis.
- Watch responsiveness, not just crashes. Android ANRs and iOS hangs (main thread blocked > ~250ms) hurt more users than crashes and are store-ranking inputs. Instrument startup time (cold and warm), frame drops on the key scrolling surfaces, and the p95 of your three business-critical flows.
- Ship breadcrumbs, scrub PII. Navigation events, network failures, and flag states attached to each crash report reconstruct the path into the failure. Strip tokens, emails, and free-text user content at the SDK layer; observability must not become a data-leak vector.
- Segment before you debug. Every metric sliced by app version, OS version, device class, and network type; most "mystery" regressions are one OS release or one low-memory device tier. An on-demand log-level bump via remote flag turns a reproducing user into a trace without a new build.
- Alert on releases, not noise. Page on: new crash signature trending in the latest version, crash-free rate crossing the floor, ANR rate above the store's bad-behavior threshold. Everything else is a dashboard reviewed at rollout checkpoints.
Boundaries
- Session replay and screen recording carry consent and privacy weight; involve legal review before enabling them, and never in auth or payment flows.
- Offline and low-end devices delay telemetry; judge a rollout on 24h of data, not the first hour.
- Observability tells you what broke, not why users leave; product analytics is a separate discipline with separate consent.