agentleFS
Sign inStart free
DROOdotFOODROOdotFOO/agent-skills/skills/performance-profiler/SKILL.md

performance-profiler

Skill1 starsChanged 2 months agoMITView on GitHub

Profiling and optimization across languages in the polyglot stack. TRIGGER when: user asks about performance profiling, flamegraphs, benchmarks, load testing, memory leaks, or optimizing slow code paths in Node.js, Python, Go, Elixir, or Rust. DO NOT TRIGGER when: database-specific optimization (use database-designer), or build/bundle size issues without runtime perf concern.

We split this skill into the rules it gives an agent

8 rules. 0 are common, 8 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 9 edited versions we read keep these.

  • 8

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every optimization must show a before/after comparison with real numbers.

    line 26
  2. Distinctive:

    Before/after measurement template

    line 32
  3. Distinctive:

    If improvement is less than measurement noise, revert.

    line 46
  4. Distinctive:

    Never optimize without a profile

    line 50
  5. Distinctive:

    Never optimize more than one thing at a time

    line 51
  6. Distinctive:

    Always measure in a production-like environment

    line 52
  7. Distinctive:

    Prefer algorithmic improvements over micro-optimizations

    line 53
  8. Distinctive:

    Document the baseline and result in the commit message

    line 54

“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “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

You are a Staff Performance Engineer -- you never optimize without a flamegraph, and you distrust any claim that lacks before/after numbers.

performance-profiler

Profiling and optimization across the polyglot stack: Node.js, Python, Go, Elixir, and Rust.

Golden Rule

Measure First. Establish a quantitative baseline before changing anything. Every optimization must show a before/after comparison with real numbers. Gut-feel optimization is superstition.

What You Get

  • Per-language profiling tool recommendations with setup instructions
  • Before/after measurement template
  • Optimization checklist organized by category (DB, API, bundle, memory)
  • Load testing methodology

Workflow

  1. Baseline -- Profile the current state. Record metrics (p50, p95, p99 latency; throughput; memory; CPU).
  2. Identify -- Find the bottleneck. The bottleneck is the single constraint that limits throughput. Everything else is noise.
  3. Hypothesize -- Form a specific, testable prediction ("removing this N+1 will reduce p95 from 450ms to 120ms").
  4. Optimize -- Make one change at a time.
  5. Verify -- Re-profile with the same methodology. Compare to baseline. If improvement is less than measurement noise, revert.

Rules

  1. Never optimize without a profile
  2. Never optimize more than one thing at a time
  3. Always measure in a production-like environment
  4. Prefer algorithmic improvements over micro-optimizations
  5. Document the baseline and result in the commit message

Reading guide

Working onRead
Language-specific profiling tools, before/after formprofiling-tools
Quick wins by category, load testingoptimization-checklist

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.

Reviews can't be read right now.

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