review-tool-speed

name	review-tool-speed
description	Review query tools for runtime optimization opportunities
user-invocable	true
disable-model-invocation	true

Enter planning mode. Investigate query tool runtime and find optimization opportunities.

The query tools (query_*.ps1 in tools/) are critical for keeping context bloat down and getting programmatic answers.

1. Investigate each query tool for potential optimizations to decrease runtime. Optimizations must be internal only — no change to output, function, or contract of the tool.
2. For each recommendation, include an analysis statement explaining why it's internal-only and doesn't change the tool's job or contract.
3. Consider: caching, pipeline vs loop, regex precompilation, redundant file reads, shared parsing, startup overhead, etc.

Research the query tools deeply. Write a plan for all optimizations found. Ignore any existing plans — create a fresh one.

Measurement & Verification Tools

Two tools in tools/ support this workflow. Use them — do not reinvent timing or golden-output infrastructure.

tools/bench_query.ps1 — Timing benchmark

Runs all query tools with representative args, reports external and internal timing.

powershell -File tools/bench_query.ps1                  # 3 iterations, full report
powershell -File tools/bench_query.ps1 -InternalOnly    # internal timing only
powershell -File tools/bench_query.ps1 -Iterations 5    # more iterations for stability

tools/verify_query.ps1 — Golden output capture & verification

Captures tool output before changes, verifies output is unchanged after. Strips timing lines automatically. 22 test cases across all query tools.

powershell -File tools/verify_query.ps1 -Capture        # save golden (prints dir path)
powershell -File tools/verify_query.ps1 -Verify <dir>   # compare against golden

Required workflow

1. Before changes: Run verify_query.ps1 -Capture and bench_query.ps1 for baseline
2. After each change: Run verify_query.ps1 -Verify <golden-dir> — must pass
3. After all changes: Run bench_query.ps1 — compare against baseline, no regressions