Agent skill · data analytics · thienanblog
performance-optimization
Coordinator-routed specialist for measured latency, CPU, memory, query, payload, rendering, bundle, caching, or build/test bottlenecks. Use after project-development-mindset establishes performance as the primary work, or directly when explicitly invoked or installed standalone. Use debugging first for unexplained failures; do not use for routine performance-aware implementation.
Why this skill is useful
Provides domain-specific knowledge on performance bottlenecks and optimization techniques that the AI wouldn't reliably generate on its own.
What it needs
About 3k tokens when loaded. Last updated 2026-08-05. 63 stars on the source repository.
What this skill does
Performance Optimization Use this skill when performance is the main concern. Measure first, optimize the confirmed bottleneck, and verify improvement without changing business behavior accidentally. Run this skill in the main conversation. Do not spawn subagents, agent teams, or delegated parallel workers unless the user explicitly approves the proposed count and scope after being told that doing so can increase usage. Ask again before expanding an approved scope. Operating Rules Do not optimize blindly. Capture a baseline or concrete symptom first. Define a benchmark envelope before comparing results: workload, starting data, cache state, command and flags, resource limits, and concurrent activity. Read project docs, architecture notes, caching rules, database rules, design-system rules, and existing performance conventions. Preserve business logic and data correctness. Prefer low-risk local improvements before broad architecture changes. Treat caching as a contract: define invalidation, freshness, and user-specific data boundaries. Treat infrastructure health as part of correctness. Reject measurements with crashes, OOM kills, unexpected restarts, failed cleanup, or orphan processes. Avoid adding dependencies or infrastructure unless measurement justifies them. If the issue is actually a bug or regression with unclear cause, return routing control to project-development-mindset and replace this workflow with debugging-workflow when available. Keep benchmarks, regression checks, and browser measurements inside this workflow when they support performance work. Route to testing-verification only if test or QA design becomes the primary deliverable. Workflow 1. Define The Performance Claim Identify what is slow, where, for whom, and compared to what. Capture baseline evidence: timing, query count, payload size, memory, CPU, bundle size, Web Vitals, screenshot, profile, or logs. Identify the environment and data size used for measurement. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills thienanblog/performance-optimization