# Performance

> Profile and optimize latency, throughput, memory, context usage, and startup costs using measured bottlenecks.

- Skill: `nafeeur/performance` (Agent Skill)
- Install (CLI): `npx skillmds@latest add nafeeur/performance`
- Raw SKILL.md: https://api.skillmd.com/api/skills/nafeeur/performance/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: nafeeur (https://skillmd.com/u/nafeeur)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/nafeeur/performance

---


# Performance Engineering

- Define the target metric, workload, baseline, and acceptable regression budget.
- Profile before optimizing; distinguish CPU, I/O, network, allocation, serialization, model latency, and context-size costs.
- Optimize the dominant path and preserve correctness under concurrency and cancellation.
- For agent systems, bound tool output, context injection, retries, and fan-out. Cache immutable discovery results.
- Add a benchmark or telemetry counter that prevents silent regression.
- Report measured before/after results and environmental limits.