# perf-theory-tester

A skill that keeps perf experiments honest: one change per hypothesis, repeated runs, a recorded verdict. Useful when you need proof that an optimization helped.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode · 1500 installs · ★ 1965
- Author: composio-community
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/composio-community/awesome-claude-plugins/tree/HEAD/perf/skills/theory-tester/SKILL.md
- Updated: 2026-07-26
- HTML: https://affpapa.org/en/nevibecode/skills/perf-theory-tester

## What it does

The skill walks the agent through a controlled experiment. It starts by confirming the baseline is clean. Then it applies a single change tied to the hypothesis and runs at least two validation passes. The code goes back to baseline before the next experiment begins.

Output follows a fixed shape: hypothesis id, a short summary of the change, metric deltas, a verdict of `accept`, `reject` or `inconclusive`, plus evidence — the benchmark command and the files touched. The skill treats `docs/perf-requirements.md` as the canonical contract to follow.

The rules are strict on purpose. One change per experiment. No parallel benchmarks. Every run leaves a record. Without that discipline, measurement noise passes for a speedup.

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add composio-community/awesome-claude-plugins --skill perf-theory-tester -a claude-code
```
```bash
npx skills add composio-community/awesome-claude-plugins --skill theory-tester -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/theory-tester/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/theory-tester/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add composio-community/awesome-claude-plugins --skill theory-tester -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/theory-tester/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add composio-community/awesome-claude-plugins --skill theory-tester -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/theory-tester/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install composio-community/awesome-claude-plugins theory-tester
```

2. **Manually: put the files in place**
- File: `.github/skills/theory-tester/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add composio-community/awesome-claude-plugins --skill theory-tester -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/composio-community/awesome-claude-plugins --path perf/skills/theory-tester
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/theory-tester/SKILL.md` — The legacy .windsurf/skills/ path still works

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add composio-community/awesome-claude-plugins --skill theory-tester -a opencode
```

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/composio-community/awesome-claude-plugins/tree/HEAD/perf/skills/theory-tester/SKILL.md

## When to use it
- prove that a code change actually made things faster
- isolate the effect of one optimization from surrounding edits
- keep a record of the benchmark command and files for each run
- get a clear accept or reject instead of a guess
- reset the tree to baseline before the next attempt

## Example prompts
- Run a controlled experiment on the hypothesis that caching cuts response time, and format the result the way the skill expects
- Take a baseline of the `/api/search` endpoint, apply one change, then give me a verdict with the commands and changed files
- Check my experiment against `docs/perf-requirements.md` and tell me what is missing before a verdict

## Similar
- [improve-codebase-architecture](https://affpapa.org/en/nevibecode/skills/improve-codebase-architecture.md) — skill · 1163035 installs
- [design-taste-frontend](https://affpapa.org/en/nevibecode/skills/design-taste-frontend.md) — skill · 678700 installs
- [tdd](https://affpapa.org/en/nevibecode/skills/tdd.md) — skill · 1141062 installs
- [skill-creator](https://affpapa.org/en/nevibecode/skills/skill-creator.md) — skill · 414180 installs
- [agent-browser](https://affpapa.org/en/nevibecode/skills/vercel-labs-agent-browser.md) — skill · 1200000 installs
- [prototype](https://affpapa.org/en/nevibecode/skills/prototype.md) — skill · 986201 installs

From the same set: [composio-community/awesome-claude-plugins](https://affpapa.org/en/nevibecode/by/composio-community/awesome-claude-plugins.md) — 8 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md