performance-budgets
Bundle, latency, and query budgets with measure-first optimization. Use when setting performance targets or fixing slowness.
- License 许可
- MIT
- 最近更新
- 2026-10-20
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- performance-budgets
- license
- MIT
- description
- Bundle, latency, and query budgets with measure-first optimization. Use when setting performance targets or fixing slowness.
- compatibility
- opencode
03 SKILL.md 正文
What I do
- Replace "make it fast" with numbers, then spend effort only where measured.
When to use me
Use when setting performance targets or diagnosing slowness.
Rules
- Budgets first: client bundle size, p95 API latency, and per-screen query counts get explicit numbers before optimization begins.
- Measure, then cut: profile (bundle analyzer, query plans, flame/path timing) before changing code; re-measure after and paste before/after numbers in TASK_PROGRESS.md.
- Usual suspects: N+1 queries, unbounded lists, render loops, unindexed filters, oversized images/assets, blocking startup work.
- Guardrails: budgets enforced in CI (bundle check, slow-test threshold) so regressions fail the build, not the user.
04 同仓库更多 Skills
demo-skill
暂无简介
web-accessibility
WCAG 2.2 AA practice for keyboard, screen reader, contrast, and motion. Use when building or auditing user interfaces.
ui-conventions
Black/white token system, accessible components, and copy rules for frontend work. Use when building screens, components, or design tokens.
test-evidence
stack-toolchain test matrices, concurrency cases, and TASK_PROGRESS proof format for QA verification. Use when writing tests or proving deliverables work.
test-driven-development
RED-GREEN-REFACTOR cycle for new behavior. Use when implementing any feature, fix, or endpoint before writing production code.
systematic-debugging
Reproduce-isolate-fix-verify loop for defects and flaky behavior. Use when facing a bug, test failure, or unexpected behavior.