agentleFS
Sign inStart free
krutikJainkrutikJain/android-agent-skills/.github/skills/android-performance-observability/SKILL.md

android-performance-observability

Skill14 starsChanged 8 months agoMITView on GitHub

Measure startup, rendering, memory, jank, vitals, logs, and crash signals for Android apps with actionable traces.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 5 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when the request is about: android performance profiling, baseline profile or macrobenchmark, app startup issue android.

    line 17
  2. Very specific:

    Read references/patterns.md when you need the measurement ladder for startup, jank, traces, and production signals.

    line 19
  3. Very specific:

    Read references/scenarios.md for repeatable profiling and trace-oriented entry points.

    line 20
  4. Distinctive:

    Treat benchmarks, traces, and vitals as different evidence sources with different noise profiles.

    line 33
  5. Distinctive:

    Prefer reproducible release-build measurements over debug-build intuition.

    line 34
  6. Distinctive:

    Keep the profiling setup stable enough that regressions are attributable to code changes instead of device or environment churn.

    line 36
  7. Distinctive:

    Before/after comparisons are explicit for any recommended optimization.

    line 66

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Android Performance Observability

When To Use

  • Use this skill when the request is about: android performance profiling, baseline profile or macrobenchmark, app startup issue android.
  • Primary outcome: Measure startup, rendering, memory, jank, vitals, logs, and crash signals for Android apps with actionable traces.
  • Read references/patterns.md when you need the measurement ladder for startup, jank, traces, and production signals.
  • Read references/scenarios.md for repeatable profiling and trace-oriented entry points.
  • Handoff skills when the scope expands:
  • android-compose-performance
  • android-ci-cd-release-playstore

Workflow

  1. Classify the symptom before choosing tools: cold start, warm start, frame/jank, scrolling, memory, ANR, crash, battery, or production vitals drift.
  2. Measure on release-like builds and physical devices whenever possible; avoid debugging from debug-only traces or profile-unfriendly builds.
  3. Pick the smallest tool that answers the question: Macrobenchmark for startup/scroll numbers, Baseline Profiles for ahead-of-time optimization, Perfetto/System Tracing for deep traces, JankStats or FrameMetrics for frame quality, and Play Vitals for field evidence.
  4. Change one thing at a time, then compare before and after traces or benchmark outputs instead of stacking multiple optimizations blindly.
  5. Hand off UI-specific rendering changes or release rollouts only after the measurement surface is stable and the bottleneck is evidenced.

Guardrails

  • Treat benchmarks, traces, and vitals as different evidence sources with different noise profiles; do not mix them casually.
  • Prefer reproducible release-build measurements over debug-build intuition.
  • Tie optimizations back to user-facing metrics such as startup time, frame pacing, ANRs, or battery impact.
  • Keep the profiling setup stable enough that regressions are attributable to code changes instead of device or environment churn.

Anti-Patterns

  • Chasing micro-optimizations before identifying whether the problem is startup, rendering, I/O, or field reliability.
  • Reading one noisy trace and presenting the result as settled fact.
  • Measuring debug builds and assuming the same behavior in production.
  • Adding Baseline Profiles or macrobenchmarks without checking whether the target path is stable enough to compare.

Review Focus

  • Startup: cold and warm launch, expensive initialization, and Baseline Profile coverage.
  • Rendering: jank, skipped frames, Compose or View invalidation churn, and long main-thread work.
  • Memory and reliability: allocations, leaks, ANRs, crashes, and Play Vitals trends.
  • Evidence quality: repeatable commands, release-like variants, and documented before/after comparisons.

Examples

Happy path

  • Scenario: Audit the repo for profiling surfaces and benchmark-related hooks before proposing a measurement plan.
  • Command: rg -n "baseline|macrobenchmark|profileable|JankStats|Trace|Perfetto" .

Edge case

  • Scenario: Keep startup and rendering investigations grounded in repeatable release-like builds.
  • Command: cd examples/orbittasks-compose && ./gradlew :app:assembleDebug

Failure recovery

  • Scenario: Keep observability requests distinct from Compose-only tuning or release automation.
  • Command: python3 scripts/eval_triggers.py --skill android-performance-observability

Done Checklist

  • The bottleneck is classified and tied to an evidence source that can be re-run.
  • The chosen tools match the symptom instead of duplicating noisy measurements.
  • Before/after comparisons are explicit for any recommended optimization.
  • UI-only or release-only work is separated from measurement and tracing.

Official References

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free