Skills2026年4月4日·1 分钟阅读

Claude Token Efficient — Drop-in CLAUDE.md for Terse Output

A single CLAUDE.md file that reduces Claude Code output verbosity by 63%. Includes 5 specialized profiles for coding, benchmarks, agents, and analysis workflows.

AI Open Source
AI Open Source · Community
Agent 就绪

先审查再安装

这个资产需要先审查。复制的指令会要求 Agent dry-run、列出写入项，确认后再继续。

Needs Confirmation · 64/100策略：需确认
Agent 入口
任意 MCP/CLI Agent
类型
Skill
安装
Single
信任
信任等级：Established
入口
Claude Token Efficient — Drop-in CLAUDE.md for Terse Output
先审查命令
npx -y tokrepo@latest install 5bfc4f12-c3d3-45c1-b973-19f51116ad7c --target codex

先 dry-run，确认写入项后再运行此命令。

TL;DR
A single CLAUDE.md file that reduces Claude Code output verbosity by 63% with 5 specialized workflow profiles.
§01

What it is

Claude Token Efficient is a CLAUDE.md configuration file that reduces Claude Code output verbosity by 63%. It includes 5 specialized profiles for different workflows: coding, benchmarks, agents, and analysis. You drop it into your project root and Claude Code automatically follows the terse output instructions.

It is designed for developers who want Claude Code to be concise and efficient rather than verbose, saving tokens and reducing noise in the output.

§02

How it saves time or tokens

The primary purpose is token reduction. By instructing Claude Code to prefer editing over rewriting, avoid re-reading files, and be concise in output, the CLAUDE.md reduces token consumption by approximately 63% according to the author's benchmarks. Less verbose output also means faster reading and less scrolling for developers.

§03

How to use

  1. Download the CLAUDE.md file to your project root
  2. Claude Code automatically reads CLAUDE.md when starting a session
  3. Choose a profile by mentioning it in your prompt, or use the default terse mode
§04

Example

# Download and add to your project root
curl -o CLAUDE.md https://raw.githubusercontent.com/drona23/claude-token-efficient/main/CLAUDE.md

The CLAUDE.md includes directives like:

§05

Approach

  • Think before acting. Read existing files before writing code.
  • Be concise in output but thorough in reasoning.
  • Prefer editing over rewriting whole files.
  • Do not re-read files you have already read unless the file may have changed.
  • Use precise, minimal diffs when editing.
§06

Output Rules

  • No preamble. No summary unless asked.
  • Code only when code is the answer.
  • One-line answers when a one-line answer suffices.
§07

Related on TokRepo

§08

Common pitfalls

  • The terse mode can be too concise for complex tasks where you need Claude to explain its reasoning; switch to the analysis profile for those cases
  • If your project already has a CLAUDE.md, you need to merge the directives rather than overwriting
  • The 63% reduction claim is based on the author's benchmarks and may vary depending on your task types

常见问题

Does this work with all Claude Code versions?+

Yes. CLAUDE.md is a standard Claude Code feature. Any version of Claude Code that reads project-level CLAUDE.md files will pick up these directives automatically when you start a session in the project directory.

Can I customize the profiles?+

Yes. The CLAUDE.md is a plain Markdown file. Edit the directives, add new profiles, or remove ones you do not need. The file is self-documenting with comments explaining each section.

Does it affect code quality?+

No. The directives control output verbosity, not reasoning depth. Claude Code still thinks thoroughly but expresses results concisely. Code quality and correctness are not impacted.

How do I switch between profiles?+

Mention the profile name in your prompt, such as 'use the benchmarks profile for this task'. The CLAUDE.md defines profile-specific behavior that activates based on context.

Can I use this with other AI tools?+

CLAUDE.md is specific to Claude Code. Other tools like Cursor use different configuration mechanisms. However, you can adapt the principles to .cursorrules or similar tool-specific config files.

引用来源 (3)
TokRepo 相关
AI Coding ToolsPrompt LibraryFeatured Workflows
🙏

来源与感谢

Created by Drona Gangarapu. Licensed under MIT.

claude-token-efficient — ⭐ 3,100+

讨论

登录后参与讨论。
还没有评论，来写第一条吧。

相关资产

Claude Code CLAUDE.md — Best Practices Template

Production-tested CLAUDE.md template for Claude Code projects. Covers coding conventions, test requirements, git workflow, and project-specific AI instructions.

Skills
Skill Factory

caveman — Token-Saving Claude Code Skill

caveman is a Claude Code skill that compresses replies and cuts output tokens ~65% on average. Install via curl/PowerShell; toggle /caveman modes.

Skills
Skill Factory

ccusage — Real-Time Token Cost Tracker for Claude Code

CLI that reads ~/.claude logs and breaks down Claude Code token spend by day, session, and project — pluggable into your statusline.

SkillsCLI Tools
Skill Factory

CC Status Board — Smart Status Bar for Claude Code

Add a context meter, AI asset discovery, and session info to your Claude Code status bar. Scans 300+ installed assets (skills, agents, MCP, plugins) and surfaces the most relevant ones as you type. Zero token cost, 100% local.

Skills
henuwangkai
◈首页🔍搜索👤我的