jfr-profiling

Installation
SKILL.md

JFR Profiling for Java/Kotlin Projects

End-to-end workflow: Gradle task → JFR → collapsed stacks (primary analysis format) → flamegraph (visualization).

Format Strategy

Format Size vs JFR Best for Command
Collapsed stacks ~1-2% LLM analysis, CI diffs, awk parsing jfrconv -o collapsed
HTML flamegraph ~5-10% Interactive human exploration jfrconv (default)
Differential flamegraph ~5-10% A/B comparison jfrconv --diff
Raw JFR 100% Archive, future re-analysis

Use collapsed stacks for LLM analysis. Format: frame1;frame2;leaf N — one line per unique call stack, count at end. No decoder needed.


Step 1 — Gradle Task Setup

Installs
1
GitHub Stars
8
First Seen
Jun 3, 2026
jfr-profiling — tstapler/dotfiles