profiling-code

Collection of examples and links that uses different profiling tools to show memory usage and timings.

Shows how to use the profiler and what the output looks like. Includes only tools that are maintained.

Installing the agent skills safely

npx skills add rudolfolah/profiling-code --all installs every skill, including the advanced Python skills that document process attachment and privileged DTrace workflows. Security scanners therefore rate those skills more cautiously than skills that only explain an interactive profiler. The rating is a prompt-risk assessment, not a report of a vulnerable npm dependency; review the skill content and grant only the capabilities that fit your environment.

Prefer installing only the skills you need rather than using --all. To see the current CLI's selection syntax, run npx skills add rudolfolah/profiling-code --help, then select skills by name. The lower-risk starting set is:

• browser-javascript-profiling
• game-profiling
• gpu-profiling
• javascript-profiling
• jvm-profiling
• python-memory-profiling
• react-profiling
• systems-profiling

Add python-cpu-profiling only when external process sampling is needed, and add python-dtrace-profiling only on a controlled macOS host where privileged DTrace access is acceptable. python-profiling is an overview that links to both advanced workflows, so review those referenced skills before enabling it. Pin or inspect the repository revision before installing, re-run the assessments after updates, and never approve an elevation or process attach solely because a skill suggested it.

awesome-profiling: Great list of profiling tools, though includes some tools for Python that are no longer maintained.

⚙️ Tools and Techniques for Profiling Code

• Game Development: C#, Unity, Godot
• Node.js: JavaScript, TypeScript
• Python: examples using cProfile, guppy3, psutil, memray, pyinstrument, tracemalloc, yappi, filprofiler, DTrace
  • Python 3.12 support for the Linux perf profiler
  • pyperformance - Python Performance Benchmarking Suite
• gprof: GNU Profiler
• Tracy Profiler: supports C, C++, Lua, Python with 3rd party bindings for Rust, Zig, C#
• valgrind
  • memcheck: memory error detection
  • cachegrind: CPU cache profiling
  • callgrind is cachegrind with a call graph and branch prediction profiler
  • massif for heap profiling
• Apple Instruments
• VisualVM: Java profiling tool
• Java Flight Recorder and JDK Mission Control: Java profiling tool
• Chrome DevTools, JavaScript in the Web Browser
• AI: Claude Skill "performance profiler" for various performance profiling tools

🧰 GPU Profiling Tools

• NVIDIA Visual Profiler
• NVIDIA Nsight Systems
• NVIDIA Nsight Compute
• Android GPU Inspector (AGI)
• Metal debugger
• AMD uProf
• Intel VTune Profiler

📚 Learning Resources

📖 Books

• "BPF Performance Tools", Brendan Gregg
• "Systems Performance: Enterprise and the Cloud, 2nd Edition", Brendan Gregg

📰 Articles

• "All my favorite tracing tools: eBPF, QEMU, Perfetto, new ones I built and more", Tristan Hume
• "Flame Graphs: The future of performance analysis", Brendan Gregg
• "Flame Graphs", Brendan Gregg
• "Sampling v. tracing", Dan Luu
• Memory and CPU profiling, "The Hacker's Guide to Scaling Python", Julien Danjou
• ebpf-apps: sample apps for eBPF for C, C++, Python, Go, and Rust.

📺 Videos

• Python profiling and performance tuning in production
• Memray: The endgame Python memory profiler
• GopherCon 2021: Go Profiling and Observability from Scratch
• eBPF
  • A Beginner's Guide to eBPF Programming with Go
  • eBPF: Fueling New Flame Graphs & more
  • Kernel Analysis Using eBPF

🤖 Agent Skills

The repository bundles the following agent skills in .claude/skills/:

Skill directory	Purpose	Notable bundled reference files
browser-javascript-profiling	Diagnose page-load cost, interaction latency, rendering jank, long tasks, and browser-retained objects with Chrome DevTools.	—
game-profiling	Profile Unity and Godot frame hitches, frame rate, allocations, memory, and rendering cost.	—
gpu-profiling	Select and use vendor and platform tools for GPU timelines, kernels, graphics captures, and counters.	nvidia.md, amd.md, intel.md, android.md, metal.md
javascript-profiling	Profile Node.js CPU use, allocations, retained heap, garbage collection, and event-loop latency.	—
jvm-profiling	Investigate JVM CPU, allocation, garbage collection, locks, safepoints, and memory with JFR, JMC, and related tools.	—
python-cpu-profiling	Compare and apply Python CPU and runtime profilers for call timing, sampling, and benchmarking.	cprofile.md, py-spy.md, pyinstrument.md, pyperformance.md, yappi.md
python-dtrace-profiling	Trace CPython function and source-line events with DTrace on macOS.	—
python-memory-profiling	Investigate Python allocations, object retention, heap growth, and process memory.	—
python-profiling	Choose an appropriate Python CPU, wall-clock, memory, or interpreter-event profiling workflow.	—
react-profiling	Measure React renders, commit duration, re-render frequency, and browser-trace correlations.	—
systems-profiling	Profile native programs with gprof, Linux perf, Tracy, Valgrind, and Apple Instruments.	gprof.md, linux-perf.md, tracy.md, valgrind.md, apple-instruments.md

Install the skills from the public GitHub repository with the skills CLI:

npx skills add rudolfolah/profiling-code

To install only one skill, use the CLI's --skill selection option:

npx skills add rudolfolah/profiling-code --skill gpu-profiling

The published collection is listed on the profiling-code skills.sh registry page.