argent-native-profiler

Profiles native iOS and Android apps to diagnose CPU hotspots, UI hangs, and memory leaks.

Updated Aug 29, 2026
One-click install
npx skills add https://github.com/Seyamalam/pstu_final --skill argent-native-profiler-seyamalam
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: argent-native-profiler
Source: https://github.com/Seyamalam/pstu_final/tree/main/.agents/skills/argent-native-profiler
Command: npx skills add https://github.com/Seyamalam/pstu_final --skill argent-native-profiler-seyamalam

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Diagnosing native-level performance problems in mobile apps is difficult without the right tooling. This Skill captures and analyzes native traces from iOS and Android devices, surfacing CPU hotspots, UI hangs, and memory leaks with severity-ranked findings and drill-down stack queries. ## Core Features & Use Cases - Cross-platform native profiling: Records iOS traces via xctrace (Instruments) and Android traces via Perfetto with an in-process WASM trace processor. - Structured bottleneck analysis: Categorizes findings as CPU hotspots, UI hangs, or memory leaks, ranked RED/YELLOW by severity, with per-hang jank reasons and blocked-function attribution on Android. - Deep stack investigation: Drill into hang stacks, function callers, thread breakdowns, and leak stacks, and reload previous trace sessions for before/after comparisons after fixes. - Use Case: A user reports the app freezes when opening a large list. Start a profiling session, reproduce the interaction, stop and analyze the trace, then query the hang stacks to find the blocking function on the main thread. ## Quick Start Profile the running app on my connected Android device and show me any UI hangs or CPU hotspots.

Frequently Asked Questions about argent-native-profiler

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I profile CPU usage and UI hangs on an Android app?

Start a Perfetto recording on the device, reproduce the slow interaction, then stop and analyze the trace. The analysis reports CPU hotspots and UI hangs with per-hang jank reasons, main-thread state breakdowns, and blocked-function attribution.

How do I find memory leaks in an iOS app with Instruments?

Record with xctrace using the Leaks instrument, then analyze the exported data for leak stacks filtered by object type. If leaks come back unattributed, re-run with malloc stack logging enabled at launch to capture real allocation backtraces.

Does Android profiling require a debuggable app build?

Yes. The target app must be debuggable or include the profileable manifest flag with android:shell set to true, otherwise perf_sample callstacks cannot be captured by Perfetto.

Why do my iOS leak results show Call stack limit reached?

Attaching xctrace to a running app has no malloc-stack history, so leaks lack allocation backtraces and appear unattributed. Cold-launch the app with malloc stack logging enabled to get responsible frames and libraries.

Can I compare profiling results before and after a performance fix?

Yes. Reload previous trace sessions from disk, re-parse them, and compare findings against the new run. Treat only consistent directional changes across two or more runs or deltas above 15 percent as actionable signal.

Is simulator profiling accurate for real device performance?

No. Simulator profiling reflects host Mac performance rather than real device hardware. Use a physical device for accurate CPU timings and memory behavior, and discount profiler-overhead entries like Hermes internals in hotspot results.