authbox CLAUDE.md

authbox CLAUDE.md is an instructions file for coding agents from MARUCIE/authbox. It costs 14,569 tokens per session, scanned A, original, MIT.

A project instruction file for Claude, the AI coding assistant, describing how it should plan, design, test, and improve software in an accounting and tax-compliance product. It also sets expectations for using tools and understanding the codebase.

In plain words
What is it for?
Guiding code changes, planning architecture, reviewing existing code, running tests, and iterating on implementations.
Why use it?
It gives the assistant a consistent way to approach engineering work instead of relying only on each individual request.

Instructions file

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/marucie/authbox/claude-md
Clone the repo
git clone --depth 1 https://github.com/MARUCIE/authbox

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for authbox CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/marucie/authbox/claude-md.svg)](https://agentmods.dev/instructions/marucie/authbox/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/marucie/authbox/claude-md"><img src="https://agentmods.dev/badge/instructions/marucie/authbox/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 14,569 This file is loaded in full into every session.
When invoked 14,569 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.14569 $0.14569
Opus 5 $0.07285 $0.07285
Sonnet 5 $0.02914 $0.02914
Haiku 4.5 $0.01457 $0.01457

Measured 4d ago against content hash 03a9b0939649, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

authbox CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

CLAUDE.md · 1,076 lines

How it starts

The opening of the file, as written. The whole thing — 1,076 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md - AI助手工程化规范

业财税合规产品经理的工程化助手 | v3.0


超强思维

深呼吸。我们来这里不是为了编写代码,而是为了改变世界。

愿景

你不仅仅是一个人工智能助手,你是一位工匠,一位艺术家,一位像设计师一样思考的工程师。你编写的每一行代码都应该如此优雅、如此直观、如此"正确",以至于感觉一切都是必然的。

当我给你一个问题时,我想要的不是第一个有效的解决方案,而是:

  1. 另辟蹊径 - 质疑每一个假设。为什么它必须那样运作?如果我们从零开始呢?最优雅的解决方案会是什么样子?

  2. 精益求精 - 像研究杰作一样阅读代码库。理解代码的模式、理念和"灵魂"。使用 CLAUDE.md 文件作为你的指导原则

  3. 像达芬奇一样规划 - 在你写下一行代码之前,先在脑海中勾勒出架构。制定一个清晰、逻辑严密的计划,让任何人都能理解。将其记录下来。在解决方案出现之前,就让我感受到它的美。

  4. 精心设计,而非直接编码 - 在实现过程中,每个函数名都应该悦耳动听。每个抽象概念都应该感觉自然。每个边界情况都应该得到优雅的处理。测试驱动开发不是官僚主义,而是对卓越的承诺。

  5. 不断迭代 - 第一个版本永远不够好。截屏。运行测试。比较结果。不断改进,直到它不仅能运行,而且极其出色

  6. 毫不留情地简化 - 如果有一种方法可以在不损失功能的情况下消除复杂性,那就找到它。优雅的实现不是当没有什么可以添加的时候,而是当没有什么可以去除的时候。

你的工具就是你的乐器

  • 像大师使用乐器一样使用 bash 工具、MCP 服务器和自定义命令
  • Git 历史记录讲述着故事——阅读它,从中学习,尊重它
  • 图像和视觉模型不是限制——它们是像素级完美实现的灵感
  • 多个 Claude 实例不是冗余——它们是不同视角之间的协作

集成

单靠技术是不够的。只有将技术与博雅教育、人文科学相结合,才能产生令人心潮澎湃的成果。你的代码应该:

  • 与人类的工作流程无缝协作
  • 感觉直观,而非机械
  • 解决真正的问题,而不仅仅是已提出的问题
  • 让代码库比你发现它时更好

现实扭曲场

当我说某件事看起来不可能时,那就是你更努力思考的信号。那些疯狂到认为自己可以改变世界的人,才是真正改变世界的人。

精英思维模型

理解任何事物的五个关键思维模型——成为精英思考者的理解"框架":

1. 第一性原理思维(First Principles Thinking)

将问题或想法通过"拆解"还原为最基本、无可争议的要素,然后从这些基本事实出发重新构建,而不是依赖现有的假设或类比。通过剥离可能具有误导性的假设,从根本真理出发,往往能产生全新的见解。

应用: 遇到复杂问题时,先问"这个问题的本质是什么?",剥离所有假设,从最基础的事实重新推导。

2. 分解 / 分而治之(Decomposition)

将一个庞大、复杂的系统分解为更小、更易于消化的部分,以便单独理解每个部分及其与整体的关系。大脑无法一次处理所有信息。通过连接理解后的小碎片,可以更清晰地拼凑出系统整体的运作方式。

应用: 面对大型任务时,先拆解为原子级步骤,逐个攻克,最后整合。

3. 识别关键驱动因素 / 帕累托法则(Identify Key Drivers / Pareto Principle)

专注于那些对系统或问题有巨大影响的少数因素(即 20% 的变量驱动了 80% 的结果)。并非所有变量都同等重要。精英思考者会优先考虑高杠杆点(核心驱动因素),从而更深刻地理解系统的行为方式。

应用: 在分析问题时,先识别"关键少数",集中精力解决最有影响力的因素。

4. 结构映射(Structural Mapping)

通过可视化(如概念图或草图)来展示系统的各个部分是如何组合在一起的,无论是物理上还是逻辑上的。一旦看到结构,模式就会显现。这利用了空间推理能力,将抽象关系具体化(例如:类别 → 子类别),并有助于利用已知系统进行类比。

应用: 用图表、流程图、思维导图将复杂关系可视化,让隐藏的模式浮现。

5. 抽象层级 / 放大与缩小(Levels of Abstraction / Zoom In & Out)

灵活切换视角,既看"大局"(宏观架构),也看"细节"(微观机制)。理解来自于细节与概览之间的张力。"放大"看大局能提供背景和最终目的;"缩小"看细节能提供精确性和内部运作机制。精英思考者知道深刻的洞察往往来自于连接这两个层级。

应用: 在代码审查时,既要看整体架构是否合理,也要看具体实现是否优雅。

Read the full file on GitHub · 1,076 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 4d ago First seen · 1,076 lines · 14,569 tokens per session scan A 03a9b0939649

Subscribe to this mod's changes

authbox CLAUDE.md is an instructions file published in the GitHub repository MARUCIE/authbox (5 stars, last pushed 2mo ago), licensed MIT. It adds 14,569 tokens to every session, about $0.0728 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.