# Performance Hints

> Route performance reviews, optimization work, profiling investigations, API or data-structure design, allocation reduction, concurrency tuning, Protocol Buffer work, and C++ performance questions to the relevant section of Jeff Dean and Sanjay Ghemawat's canonical Abseil Performance Hints article. Use for broad or multi-area performance work when more than one focused performance subskill may apply.

- Skill: `boshjerns/performance-hints` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add boshjerns/performance-hints`
- Raw SKILL.md: https://api.skillmd.com/api/skills/boshjerns/performance-hints/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Integrations & APIs
- Author: boshjerns (https://skillmd.com/u/boshjerns)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/boshjerns/performance-hints

---


# Performance Hints Router

Use this skill as a dispatcher. Read only the focused skills relevant to the code and evidence at hand. Treat every optimization as optional until measurement or a strong cost estimate supports it.

## Source and attribution

Canonical source: [Performance Hints](https://abseil.io/fast/hints) by Jeff Dean and Sanjay Ghemawat. Original version 2023-07-27; consult the page for its current update date. This package supplies navigation only. The fetcher reads the live article and does not vendor its contents.

Suggested citation from the source: `Jeffrey Dean & Sanjay Ghemawat, Performance Hints, 2025, https://abseil.io/fast/hints.html`

## Route

- General philosophy: `performance-thinking`
- Back-of-the-envelope costs: `performance-estimation`
- Profiling and benchmarks: `performance-measurement`
- API shape: `performance-api-design`
- Algorithms: `performance-algorithms`
- Data layout and representation: `performance-memory-representation`
- Allocations and copies: `performance-allocations`
- Repeated or unnecessary work: `performance-unnecessary-work`
- Binary and instruction footprint: `performance-code-size`
- Concurrency and synchronization: `performance-parallelism`
- Protocol Buffers: `performance-protobuf`
- C++ containers and types: `performance-cpp`
- Batched transformations: `performance-bulk-operations`
- Worked changes combining techniques: `performance-case-studies`
- Authors' recommended resources: `performance-further-reading`

## Read a canonical section

Run:

```bash
python3 ~/.codex/skills/performance-hints/scripts/fetch_section.py SECTION_ID
```

The command emits the exact live HTML beginning at the selected `<h2>` and ending immediately before the next `<h2>`. Use `--list` to display valid section IDs and `--output text` for readable text.