performance-review_skill

Overview

This skill performs systematic performance analysis to locate bottlenecks, evaluate resource usage, and produce targeted optimization recommendations. It covers algorithm complexity, CPU/memory/I/O assessment, concurrency checks, and database query tuning. The goal is measurable performance improvements and practical implementation steps.

How this skill works

The skill inspects code patterns, runtime metrics, and query plans to identify hotspots and anti-patterns. It analyzes time/space complexity, reviews resource consumption traces, detects concurrency issues (locks, race conditions), and evaluates database access for N+1 queries and missing indexes. It outputs prioritized fixes, expected impact estimates, and testing/monitoring suggestions.

When to use it

• Application has high response times or inconsistent latency under load
• CPU, memory, or I/O usage frequently approaches capacity limits
• Suspected N+1 queries or slow database operations
• New features introduce concurrency or scaling regressions
• Before major releases or infrastructure cost-reduction efforts

Best practices

• Measure first: gather benchmarks and production traces before changing code
• Prioritize fixes by impact × effort and address high-frequency hotspots first
• Prefer algorithm and data-structure improvements over micro-optimizations
• Use caching and connection pooling judiciously with eviction/consistency plans
• Introduce monitoring and alerts for KPIs (P95 latency, CPU, memory, cache hit rate)

Example use cases

• Replace O(n²) loops with hash-based approaches to reduce latency for large inputs
• Batch and bulk I/O operations to lower syscall and network overhead
• Refactor N+1 database access into single JOIN or bulk query with mapping
• Add Redis caching for hotspot endpoints and define TTL/invalidation rules
• Introduce benchmark and load test suites (Benchmark.js, Artillery) before/after changes