v0.x This is the released v0.x line — the version that ships in release tarballs and is dogfooded in this repository. The v1.0 rebuild is documented separately in the v1.x preview.

Performance Skills

Skills for performance analysis, optimization, and load testing.

performance-profiling

Performance analysis methodology and profiling techniques for CPU, memory, and I/O. Flame graphs, benchmarking, and regression detection. Use when optimizing performance, profiling bottlenecks, or reviewing performance-critical code.

Triggers: When identifying bottlenecks, profiling CPU/memory/I/O, interpreting flame graphs, setting up benchmarks, or optimizing slow code paths. Tools: Bash Read Write References: profiling-tools.md

Key capabilities:

  • Follow the full performance analysis cycle: Identify, Measure, Profile, Optimize, Verify
  • CPU profiling with sampling profilers and flame graph generation
  • Memory profiling to detect leaks, allocation pressure, and unbounded growth
  • I/O profiling for disk, network, and database bottlenecks (N+1 queries, connection pooling, slow queries)
  • Benchmarking with statistical significance and regression detection
  • Flame graph interpretation: reading X-axis (alphabetical, not time), Y-axis (stack depth), and differential flame graphs
  • Common optimization patterns: algorithmic improvements, batching, caching, pooling, lazy evaluation, data layout

??? example “Example usage” Slow API endpoint: Measures end-to-end latency, profiles the handler, discovers 80% of time spent in 47 sequential database queries (N+1 problem). Rewrites as a single JOIN query, reducing response time from 3 seconds to 120ms.

caching-strategies

Caching patterns including cache-aside, write-through, TTL strategies, cache invalidation, and HTTP caching. Use when designing caching layers, optimizing response times, or debugging cache-related issues.

Triggers: When adding caching to reduce latency, choosing caching patterns, configuring HTTP caching headers, debugging stale data or cache stampede issues, or designing invalidation strategies. Tools: None References: None

Key capabilities:

  • Choose the right caching pattern: cache-aside, read-through, write-through, write-behind
  • Design TTL strategies with jitter to prevent thundering herd on expiration
  • Cache invalidation via event-driven, tag-based, or versioned key approaches
  • HTTP caching configuration: Cache-Control, ETag, CDN caching with s-maxage and Surrogate-Key
  • Prevent cache stampede with locking (mutex), probabilistic early expiration (XFetch), and stale-while-revalidate
  • Cache warming strategies for deploys and predictable access patterns

??? example “Example usage” Product page too slow: Profiles the endpoint and finds 3 database queries per request. Implements cache-aside with Redis: product data (TTL 10min), category tree (TTL 1hr), user-specific pricing (TTL 60s, private). Adds stale-while-revalidate to HTTP headers. Response time drops to 45ms on cache hit.

concurrency-patterns

Concurrency and parallelism patterns including async/await, threads, actors, channels, and deadlock prevention. Use when designing concurrent systems, debugging race conditions, or choosing between concurrency models.

Triggers: When choosing between threads, async/await, or actors; designing concurrent pipelines; debugging deadlocks or race conditions; implementing producer-consumer or fan-out/fan-in patterns. Tools: None References: patterns-catalog.md

Key capabilities:

  • Distinguish concurrency from parallelism and choose based on I/O-bound vs CPU-bound bottlenecks
  • Choose the right model: async/await, OS threads, green threads, actors, channels, or thread pools
  • Manage shared state safely with Mutex, RwLock, and atomic operations
  • Prevent deadlocks via lock ordering, try-lock with timeout, reduced lock scope, and lock-free algorithms
  • Detect and fix race conditions using ThreadSanitizer, cargo miri, and pattern recognition (TOCTOU, partial init)
  • Implement backpressure with bounded channels, rate limiting, load shedding, and reactive streams

??? example “Example usage” Go service deadlocks under load: Enables mutex profiling with GODEBUG=mutexprofile, identifies two goroutines acquiring locks on userCache and sessionCache in opposite orders. Fixes by establishing consistent lock ordering and reducing the critical section.

load-testing

Load testing methodology including test types, scenario design, and capacity planning. Use when planning load tests, analyzing test results, or setting up performance testing in CI.

Triggers: When planning or running load tests, choosing tools, designing test scenarios, analyzing results, estimating capacity, or setting up performance testing in CI pipelines. Tools: Bash Read Write References: None

Key capabilities:

  • Choose the right test type: smoke, load, stress, spike, and soak/endurance tests
  • Design realistic scenarios with user journeys, think time, traffic distribution, and authentication
  • Capture essential metrics: latency percentiles (p50/p95/p99), throughput (RPS), error rate, and resource utilization
  • Tool selection guidance: k6, Locust, Gatling, wrk, hey, vegeta
  • Identify bottlenecks from results: linear latency climb, periodic spikes, error thresholds, throughput plateaus
  • Capacity planning: find throughput ceiling, calculate headroom, estimate scaling needs
  • CI integration with performance gates and relative thresholds

??? example “Example usage” Pre-Black Friday load test: Designs a test plan with smoke test first, then load test at 2x normal traffic, then stress test at 5x to find the breaking point. Uses k6 with scenarios modeling the top 5 user journeys weighted by actual traffic distribution. Configures thresholds at p95 < 500ms and error rate < 0.1%.