Performance Deity ⚡

A performance engineering plugin for Claude Code — and a CLAUDE.md template for any AI coding assistant.

Nine benchmark-driven skills that enforce a strict, phased performance methodology. Every skill requires proof before presenting a change: a before/after table with real numbers. No guessing.

---

Install via Claude Code Plugin

/plugin marketplace add https://github.com/v0idOS/performance-deity
/plugin install performance-deity@performance-deity

This gives you nine slash commands in Claude Code:

Command	What it does
/optimize	Benchmarks, analyzes Big-O, rewrites, proves improvement
/memory	Instruments memory, proves the leak grows, fixes the reference
/concurrency	Fires 1k–10k concurrent requests, proves failure, adds atomic ops
/database	Runs EXPLAIN ANALYZE, kills N+1 loops, provides CREATE INDEX statements
/stress-test	Bombards code with extreme inputs, documents crashes, hardens the boundary
/network	Profiles payload bytes, enforces Brotli/ETag, migrates to GraphQL/protobuf
/build	Fixes Docker layer order, Webpack caching, GitHub Actions pipelines
/telemetry	Wraps code in OpenTelemetry spans, proposes concrete alert rules
/ui	Audits React re-renders, GPU-accelerates animations, virtualizes long lists

---

Or: Drop One File (No Plugin Required)

If you're not using Claude Code's plugin system, copy CLAUDE.md directly into your project root. Claude Code reads it automatically on every session — no flags, no config.

git clone https://github.com/v0idOS/performance-deity
cp performance-deity/CLAUDE.md /your/project/CLAUDE.md

For Cursor or Windsurf:

cp performance-deity/CLAUDE.md /your/project/.cursorrules

---

Global Rules (Enforced by Both Methods)

• Never suggest a code change for performance without benchmarking the existing code first.
• Every optimization must include a before/after comparison table with real numbers.
• Prefer algorithmic improvements over micro-optimizations.
• If a proposed change does not measurably beat the baseline, discard it and try a different approach.
• Every final report must explain why the change is faster, grounded in CPU architecture, memory layout, or I/O behavior.

---

Ephemeral Benchmarking (True Zero-Config)

Unlike other plugins that require you to copy a tools/ folder into your project, Performance Deity generates its own benchmarking tools on the fly.

Whenever Claude needs to prove a latency reduction, it writes an ephemeral micro-benchmark script in the target language (Node.js, Python, Rust, etc.), runs it to collect P95 and Average execution times (including a warm-up phase to prime JIT caches), and then deletes the script. No clutter, no dependencies.

---

Pre-Commit Hook (Optional)

Blocks commits containing unresolved TODO: optimize markers:

cp hooks/pre-commit .git/hooks/pre-commit
chmod +x .git/hooks/pre-commit

---

Repo Structure

.
├── CLAUDE.md                   ← Drop-in template for any AI assistant
├── README.md
├── LICENSE                     ← MIT
├── .claude-plugin/
│   ├── plugin.json             ← Plugin manifest (Claude Code plugin system)
│   └── marketplace.json        ← Marketplace catalog (buildwithclaude.com)
├── skills/                     ← Slash command skill files (Claude Code plugin)
│   ├── optimize/SKILL.md
│   ├── memory/SKILL.md
│   ├── concurrency/SKILL.md
│   ├── database/SKILL.md
│   ├── stress-test/SKILL.md
│   ├── network/SKILL.md
│   ├── build/SKILL.md
│   ├── telemetry/SKILL.md
│   └── ui/SKILL.md
├── hooks/
│   └── pre-commit              ← Optional git hook
└── archive/                    ← Original SKILL.md files (pre-refactor)

---

Roadmap

• Cross-platform benchmark runners (Python, Node.js, PowerShell, Bash)
• All nine skills in a single CLAUDE.md
• Claude Code plugin with proper marketplace.json and slash commands
• MIT license
• GitHub Actions workflow: run benchmarks on PRs, fail on P95 regression
• .cursorrules variant maintained in sync with CLAUDE.md

---

License

MIT

---

Built for developers who demand blazing-fast software.