When you're building realtime dashboards, market data feeds, or anything where p95 latency actually matters, this skill keeps Claude focused on the hot path. It pushes you to split metrics properly (p50/p95/p99, throughput, freshness age, queue depth), map the entire chain from event to user-visible state, and verify with live readbacks instead of guessing. The optimization order is pragmatic: remove round trips, cache with freshness metadata, batch calls, then move compute closer to data. It won't let you hide stale data behind fast cache hits or claim millisecond behavior without measurement, which is exactly the discipline you need when latency is user-facing.
npx -y skills add affaan-m/ecc --skill latency-critical-systems --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills