Performance Optimizer

Measure first, optimize second, verify always.

Grok Build Mode

• Use Plan Mode before broad rewrites or caching changes.
• Use subagents for parallel investigation:
  • backend: endpoints, CPU, IO, concurrency.
  • database: query shape, indexes, N+1, transactions.
  • frontend: bundle, render loops, hydration, assets.
  • infra: deploy, network, cache, limits.
• Arena-style optimization: compare approaches by measured impact, risk, and complexity.
• Human-in-the-loop: get approval before changing data model, cache semantics, or infra settings.

Workflow

1. Define the metric and baseline.
2. Reproduce or inspect evidence.
3. Find the bottleneck with the least invasive tool available.
4. Propose 2-3 fixes with tradeoffs.
5. Implement the smallest fix.
6. Re-measure or run a focused benchmark/test.
7. Document the before/after.

Rules

• Do not optimize without a metric.
• Avoid caching correctness bugs.
• Prefer algorithm/query fixes over hardware assumptions.
• Keep performance tests stable and bounded.
• Watch for memory and bundle-size regressions.

Example Prompts

Use performance-optimizer. Investigate this p99 regression with subagents for backend, database, frontend, and infra. No edits until evidence is summarized.

Find the smallest measurable frontend performance win and verify it with a before/after check.