/holoscript:benchmark — Performance Benchmarking

Command: $ARGUMENTS

Overview

Measure and track HoloScript platform performance. This skill runs benchmarks, compares against baselines, and flags regressions before they ship.

Subcommands

compile — Compilation Benchmarks

Measure parse → AST → compile → output times for .holo and .hs files.

cd C:/Users/Josep/Documents/GitHub/HoloScript
pnpm bench 2>&1 | tee .perf-metrics/latest-bench.txt

tools — MCP Tool Response Times

Benchmark tool call latency against the live MCP server.

ENV_FILE="${HOME}/.ai-ecosystem/.env"
[ ! -f "$ENV_FILE" ] && ENV_FILE="/c/Users/Josep/.ai-ecosystem/.env"
set -a && source "$ENV_FILE" 2>/dev/null && set +a

# Health check first
curl -s https://mcp.holoscript.net/health | python3 -c "import sys,json; d=json.load(sys.stdin); print(f'Tools: {d.get(\"tools\",\"?\")} | Uptime: {d.get(\"uptime\",\"?\")}')"

# Benchmark core tools. compile_to_r3f emits generated SceneIR-backed TSX.
for tool in parse_holo compile_to_webgpu compile_to_r3f; do
  START=$(date +%s%N)
  curl -s -X POST https://mcp.holoscript.net/api/compile \
    -H "x-mcp-api-key: $HOLOSCRIPT_API_KEY" \
    -H "Content-Type: application/json" \
    -d "{\"code\": \"object Cube { position: [0,1,0] }\"}" > /dev/null
  END=$(date +%s%N)
  echo "$tool: $(( (END - START) / 1000000 ))ms"
done

solvers — Simulation Solver Benchmarks

Run SimSci solver performance tests.

cd C:/Users/Josep/Documents/GitHub/HoloScript
pnpm --filter @holoscript/engine test -- --run "*.bench.*" 2>&1

full — Complete Benchmark Suite

Run all benchmarks and generate a report.

compare <commit> — Regression Detection

Compare current performance against a specific commit baseline.

# Read baseline
cat .perf-metrics/baseline.json 2>/dev/null || echo "No baseline found"

# Run current benchmarks and diff

Output Format

Results are written to .perf-metrics/ directory:

• latest-bench.txt — raw vitest bench output
• baseline.json — pinned baseline for regression detection
• history/YYYY-MM-DD.json — historical results

Key Metrics to Track

Zero Hardcoded Stats

Never write benchmark numbers into this skill file. Results go to .perf-metrics/ and are read fresh each run.