bench

If $ARGUMENTS is empty or not provided:

Usage: /bench <target>

• /bench pkg/auth/login.go — benchmark all exported functions in a file
• /bench ProcessOrder — benchmark a specific function
• /bench pkg/utils/ — benchmark all functions in a package

Workflow: detect tooling/existing benchmarks → analyze target → generate table-driven benchmarks → run with -benchmem -count=6 → compare against baseline (if benchstat) → profile CPU + memory → summarize + suggest optimizations.

Ask: "What file, function, or package would you like me to benchmark?"

---

If $ARGUMENTS is provided:

Generate and run Go benchmarks for the specified code with statistical rigor.

Loop Initialization

!if [ ! -x "${CLAUDE_PLUGIN_ROOT}/scripts/setup-loop.sh" ]; then echo "ERROR: Plugin cache stale. Run /gopher-ai-refresh (or refresh-plugins.sh) and restart Claude Code."; exit 1; else "${CLAUDE_PLUGIN_ROOT}/scripts/setup-loop.sh" "bench" "COMPLETE"; fi

Configuration

• Target: $ARGUMENTS (file, function, or package)

Steps

1. Detect Benchmark Environment

which benchstat 2>/dev/null && echo "benchstat: available" || echo "benchstat: not found"
grep -rl 'func Bench' --include='*_test.go' ./path/to/package/
ls .bench-baseline*.txt 2>/dev/null

Also detect: existing _test.go patterns, testify vs stdlib, existing .pprof files.

If existing benchmarks for the target, ask via AskUserQuestion:

Option	Action
Run existing	Run only, skip generation
Augment	Add new cases alongside existing
Generate fresh	Create new benchmarks in a separate file

2. Analyze Target

Read the target and extract: function signatures, input types/sizes (for realistic data), dependencies/interfaces (for setup), I/O operations (special handling), allocation-heavy patterns (string concat, slice append, map ops), concurrency patterns (channels, mutexes).

For each function: classify CPU- vs memory- vs I/O-bound; identify variable-size sub-benchmarks; need for b.ResetTimer(); exported vs unexported (affects test package choice).