# Optimize Application Performance with Evidence

Performance changes often add complexity without proving user impact. This skill guides measurement, focused fixes, verification, and regression protection across application layers.

## Install

```bash
npx skillstore add addyosmani/performance-optimization
```

## Metadata

- Status: approved
- Slug: addyosmani-performance-optimization
- Skillstore revision: r1
- Version status: missing
- Tree hash: bb67dc7980db00bb8b5de892b696fc98336553b0dabfe2c12cb553e87b78d60b
- Author: addyosmani
- GitHub username: addyosmani
- License: MIT
- Repository: https://github.com/addyosmani/agent-skills/tree/be4e44a9fbc5e8df0beaefadbb28bd22ee61cc39/skills/performance-optimization
- Ref: 5526951beaeca22519572a471618555839b53b05
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: scripts, external\_commands, filesystem
- Quality score: 77
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/addyosmani-performance-optimization
- Manifest: https://skillstore.pages.dev/api/skills/addyosmani-performance-optimization/manifest

## Capabilities

- Establishes repeatable baselines using synthetic measurements and real user monitoring.
- Maps frontend and backend symptoms to likely bottlenecks and appropriate profiling methods.
- Reviews N\+1 queries, pagination, indexes, connection pools, caching, images, rendering, and bundle size.
- Defines Core Web Vitals targets and practical performance budgets.
- Compares before and after results while accounting for measurement variance.
- Designs CI gates and field monitoring to detect performance regressions.

## Use Cases

- Improve Web Vitals: Diagnose poor LCP, INP, or CLS and prioritize measurable frontend changes.
- Reduce API Latency: Trace slow endpoints through queries, connection pools, caching, and application code.
- Prevent Regressions: Define budgets, repeatable benchmarks, CI gates, and production monitoring for critical user metrics.

## Prompt Templates

### Find the First Measurement

```
My application feels slow during [workflow]. Help me choose one user-facing metric, a representative test condition, and a repeatable baseline.
```

### Diagnose a Bottleneck

```
Analyze these measurements and traces for [workflow]: [evidence]. Identify the likely bottleneck, explain the evidence, and propose the smallest useful experiment.
```

### Review an Optimization

```
Review this proposed performance change: [change]. Compare baseline and result data, check correctness risks, account for variance, and recommend keep or revert.
```

### Design Regression Controls

```
Create a performance protection plan for [system]. Define budgets, benchmark conditions, sample strategy, CI enforcement, field alerts, and an investigation workflow.
```

## Limitations

- It does not provide measurements without access to the application and representative workloads.
- Example thresholds require adjustment for product requirements, infrastructure, and user conditions.
- Database recommendations require validation against real query plans and production data distributions.
- Optimization changes still require functional tests and review for correctness.

## Best Practices

- Measure the same user-facing metric before and after each isolated change.
- Validate performance gains only when functional tests remain green.
- Record successful and reverted experiments with conditions and results.

## Anti Patterns

- Do not optimize from intuition without profiling evidence.
- Do not combine several changes into one measurement.
- Do not keep complexity when improvement remains within measurement noise.

## Security Audit

- Audited at: 2026-09-13T13:03:26.303\+00:00
- Summary: All 55 static findings are false positives caused by Markdown syntax, fixed local import examples, and relative documentation links. The skill contains performance guidance and named tooling examples, with no prompt injection, arbitrary command construction, or filesystem operation.

## Stats

- Views: 0
- Downloads: 0
- Favorites: 0
- Popularity score: 0