メニュー
Establish performance baselines and detect regressions using flamegraph analysis. Use when optimizing performance-critical code, investigating performance issues, or before creating commits with performance-sensitive changes.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Run comprehensive Rust code quality checks including compilation, linting, documentation, and tests. Use after completing code changes and before creating commits.
Enforce the "Clean Imports over Inline Absolute Paths" rule by removing inline crate:: prefixes and adding proper use statements.
Rules and guidelines for creating well-formatted commit messages, including 72-char limits, scope prefixes, and trailer blocks for tasks, PR closing, and attribution.
Zero-allocation string building strategies. Use when formatting strings, generating ANSI codes, or writing hot loops to avoid heap allocations and Formatter state machine overhead.
Run clippy linting, enforce comment punctuation rules, format code with cargo fmt, and verify module organization patterns. Use after code changes and before creating commits.
Standard for writing structured tracing logs behind debug flags.
| name | analyze-performance |
| description | Establish performance baselines and detect regressions using flamegraph analysis. Use when optimizing performance-critical code, investigating performance issues, or before creating commits with performance-sensitive changes. |
Follow these steps to analyze performance and detect regressions:
Run the automated benchmark script to collect current performance data:
./run.fish run-examples-flamegraph-fold --benchmark
What this does:
tui/flamegraph-benchmark.perf-foldedImplementation details:
script-lib.fishCompare the newly generated flamegraph with the baseline:
Baseline file:
tui/flamegraph-benchmark-baseline.perf-folded
Current file:
tui/flamegraph-benchmark.perf-folded
The baseline file contains:
Compare the two flamegraph files to identify regressions or improvements:
Key metrics to analyze:
Hot path changes
Sample count changes
Call stack depth changes
New allocations or I/O
Create a comprehensive report analyzing the performance changes:
Report structure:
# Performance Regression Analysis
## Summary
[Overall performance verdict: regression, improvement, or neutral]
## Hot Path Changes
- Function X: 1500 → 2200 samples (+47%) ⚠️ REGRESSION
- Function Y: 800 → 600 samples (-25%) ✅ IMPROVEMENT
- Function Z: NEW in current (300 samples) 🔍 INVESTIGATE
## Top 5 Most Expensive Functions
### Baseline
1. render_loop: 3500 samples
2. paint_buffer: 2100 samples
3. diff_algorithm: 1800 samples
...
### Current
1. render_loop: 3600 samples (+3%)
2. paint_buffer: 2500 samples (+19%) ⚠️
3. diff_algorithm: 1700 samples (-6%) ✅
...
## Regressions Detected
[List of functions with significant increases]
## Improvements Detected
[List of functions with significant decreases]
## Recommendations
[What should be investigated or optimized]
Present the regression report to the user with:
When to update the baseline:
Only update when you've achieved a new "best" performance state:
How to update:
# Replace baseline with current
cp tui/flamegraph-benchmark.perf-folded tui/flamegraph-benchmark-baseline.perf-folded
# Commit the new baseline
git add tui/flamegraph-benchmark-baseline.perf-folded
git commit -m "perf: Update performance baseline after optimization"
See baseline-management.md for detailed guidance on when and how to update baselines.
The .perf-folded files contain stack traces with sample counts:
main;render_loop;paint_buffer;draw_cell 45
main;render_loop;diff_algorithm;compare 30
Format:
1. Make code change
↓
2. Run: ./run.fish run-examples-flamegraph-fold --benchmark
↓
3. Analyze flamegraph vs baseline
↓
4. ┌─ Performance improved?
│ ├─ YES → Update baseline, commit
│ └─ NO → Investigate regressions, optimize
└→ Repeat
For more granular performance analysis, consider:
Run benchmarks for specific functions:
cargo bench
When to use:
#[bench]Generate visual flamegraph SVG:
cargo flamegraph
When to use:
Requirements:
flamegraph crate installedFor deep investigation:
# Profile with perf
perf record -F 999 --call-graph dwarf ./target/release/app
# Generate flamegraph
perf script | stackcollapse-perf.pl | flamegraph.pl > flame.svg
When analyzing flamegraphs, watch for:
render_loop;Vec::push;alloc::grow 500 samples ⚠️
Problem: Allocating in tight loops Fix: Pre-allocate or use capacity hints
process_data;String::clone 300 samples ⚠️
Problem: Unnecessary data copies
Fix: Use references or Cow<str>
a;b;c;d;e;f;g;h;i;j;k;l;m 50 samples ⚠️
Problem: Too much abstraction or recursion Fix: Flatten, inline, or optimize
render_loop;write;syscall 200 samples ⚠️
Problem: Blocking I/O in rendering Fix: Buffer or defer I/O
After performance analysis:
This skill includes additional reference material:
baseline-management.md - Comprehensive guide on when and how to update performance baselines: when to update (after optimization, architectural changes, dependency updates, accepting trade-offs), when NOT to update (regressions, still debugging, experimental code, flaky results), step-by-step update process, baseline update checklist, reading flamegraph differences, example workflows, and common mistakes. Read this when:
check-code-quality - Run before performance analysis to ensure correctnesswrite-documentation - Document performance characteristics/check-regression - Explicitly invokes this skillperf-checker - Agent that delegates to this skilltui/*.perf-folded filesscript-lib.fishflamegraph.pl to generate SVGs