pi2dsh: Instructions file for Claude Code

CLAUDE.md

pi2dsh CLAUDE.md is an instructions file for Claude Code from weijiafu14/pi2dsh. It costs 14,464 tokens per session, scanned C, original, MIT.

Repository instructions for MiniMind Notes, a learning project with teaching modules, documentation, runnable experiments, model code, training code, and datasets. They describe its folders, package manager, documentation commands, and experiment setup.

In plain words
What is it for?
Use them when editing lessons or notes, running the VitePress documentation site, building a preview, or running Python experiments and training-related code.
Why use it?
They show where educational content, notes, experiments, and machine-learning code belong. They also provide the commands needed to run the documentation site and example programs.

Instructions file for Claude Code

Written for Claude Code: the file is CLAUDE.md. Also seen: mentions CLAUDE.md; mentions Codex.

This is weijiafu14/pi2dsh's own configuration. It tells Claude Code how to work on pi2dsh itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything pi2dsh configures →

Reuse

Borrowing it

Nothing to install: this file belongs to weijiafu14/pi2dsh. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/weijiafu14/pi2dsh/main/CLAUDE.md
Clone the repo
git clone --depth 1 https://github.com/weijiafu14/pi2dsh

Made for: Claude Code.

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 pi2dsh CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/weijiafu14/pi2dsh/claude-md.svg)](https://agentmods.dev/instructions/weijiafu14/pi2dsh/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/weijiafu14/pi2dsh/claude-md"><img src="https://agentmods.dev/badge/instructions/weijiafu14/pi2dsh/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 14,464 This file is loaded in full into every session.
When invoked 14,464 The same file — it is already loaded in full.
Security scan C 1 finding. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.14464 $0.14464
Opus 5 $0.07232 $0.07232
Sonnet 5 $0.02893 $0.02893
Haiku 4.5 $0.01446 $0.01446

Measured 9d ago against content hash de6a3912b15f, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade C, and why

pi2dsh CLAUDE.md scanned grade C with 1 finding 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 9d 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.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

55GB)。处置:`rm -rf /tmp/pi2dsh-*` + `npx pnpm@11 store prune`(只清无引用
CLAUDE.md · 614 lines

How it starts

The opening of the file, as written. The whole thing — 614 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — pi2dsh 工作准则

pi2dsh:通用 Pi Host ABI 兼容层,让 Pi 生态插件原样跑在 DeepSeek Harness (DSH) 上。本文件是本仓库一切工作的标准,任何新 session 开工前先读完。 违反任何一条 = 返工。每条标准背后的真实事故记录在 docs/STANDARDS.md——改标准前先读事故。

总纲一句话:对用户,一切是 DSH;对插件,一切是 Pi;中间层是唯一的 翻译官,且能借 DSH 官方的力就绝不自己造。

一、用户安装使用标准

用户只用 DSH 官方命令,装一次引擎,之后装谁用谁,全程没有我们发明的步骤。

  • 姿势:dsh plugin add pi2dsh 装一次引擎;之后 dsh plugin add <Pi插件原包> 直装 npm 原包。零转换、零生成产物;装了才挂、卸了就没。 代码里没有第二条路:convert/host 命令、generateBundle/generateHostBundle 导出、src/generator.ts 已整体删除(2026-08-16)。留着它们的真实代价不是 "多一个特例",而是验证会架在错的路上 —— 主集成测试和真机端到端都曾 在装转换产物,跑得再绿也证明不了用户那条路。开发和测试必须同一条路。
  • 一份引擎实例挂所有插件:一个模型目录、一个 /login、一个凭证存储、 一个升级单元。禁止多份桥拷贝各自为政(事故:/login-2 自撞)。host 级 资源(provider 目录/catalog/伴生映射/登录/凭证存储)经 SharedHostState 跨包单份共享;包级资源(tools/commands/events)各归各。零个社区 Pi 包 也必须挂 host 级运行时:内建 OAuth provider、/login、凭证恢复和伴生路由 属于引擎,不得因插件发现结果为空而跳过。零包 profile 还是最快挂载路径, 对组合服务的挂载期立即探测在这里必挂(2026-08-30 事故:凭证恢复用 optionalService 立即探测 credentials,零包时服务未组合、存量登录路由全丢 MISSING_CREDENTIAL,装任意包就"碰巧"好——对组合服务一律 ctx.inject 等到位, 且零包 profile 是必测回归形状,契约测试在 tests/dsh-runtime.spec.ts)。
  • 升级解耦:升引擎不动插件、升插件不动引擎;lockfile 锁死,只有显式 add <pkg>@latest 才动;pi2dsh inspect <pkg>@<版本> 是升级预检门。
  • 发现机制 = 读 profile 依赖清单(每项都是用户显式 add 的)+ Pi 官方 pi 字段/目录约定判定包身份;绝不扫 node_modules(Prettier 3 弃用 目录扫描的公开教训)。config packages/exclude 显式收窄兜底。
  • 引擎自身依赖必须干净:不带任何安装脚本(pnpm 对传递依赖的安装脚本 报错性拦截,用户第一条命令就会炸——事故:pi-ai→genai→protobufjs)、 不拖 CLI-only 大件(事故:typescript 23MB 白下载;现为 optional peer + 懒加载分包,改依赖后必须验证引擎 chunk 的加载路径)。
  • 撞上宿主安全门(pnpm 构建脚本审批)不绕:那是用户拍板的权利。文档 写清应对即可(allowBuilds 设 true / approve-builds)。
  • 加/卸插件后要重启 dsh(挂载在启动时);先卸插件再卸引擎;伴生路由等 引擎配置是 per-profile 的,每个用到的 profile 配一份。

二、Pi 插件处理标准(用户面界线,铁律)

插件说 Pi 话,用户说 DSH 话,中间层负责翻译——用户面前永远没有 Pi。

  • 用户接触面——要动手写的配置、要看的文档教程、要敲的命令、报错里的 指引——一律 DSH 形状、DSH 官方机制:配模型 = DSH settings 的 llm-pi-ai: 段;配伴生路由 = 引擎的 cordis 插件 config (cordis.patch.yml);凭证 = DSH credentials 引用(apiKeyEnv)。
  • Pi 形状只允许活在两处:插件视野(shim/投影/事件)与中间层内部 实现(vendored 源码、内部存储如 auth.json)。
  • 判据:用户需要亲手读写的东西里出现 Pi 词汇/格式 = 泄漏 = 返工 (事故:models.json 作为"Pi 标准配置入口"被搬进 DSH 用户世界,教 DSH 用户写 Pi 格式文件,最终全链删除)。
  • 我们兼容的对象是插件代码,不是把 Pi 生态的用户习惯搬给 DSH 用户。 "Pi 教程照搬可用"不是目标,是泄漏。
  • 插件自身配置的标准(用户怎么配好一个 Pi 插件)分三层:
    1. 环境变量(主路径):Pi 插件生态的主流配置面是 env (VISION_BRIDGE_/PI_VISION_),env 是宿主中立的——DSH 用户设 env 是纯 DSH 动作,零泄漏。examples 教这条。
    2. 插件自带斜杠命令:插件用命令管理自己的配置(/vision),命令经 中间层进 DSH 命令面板——用户敲的是 DSH 面板里的命令。
    3. 插件内部落盘:插件以为在写 Pi config 目录,实际被重定向到 $DSH_HOME/pi2dsh/ 内部目录——文件在,但不是用户接触面,任何 文档都不教用户碰它(auth.json 同理)。 判据:用户给插件配置的动作只有"设 env、敲插件命令"两种;任何"教用户 手工编辑 Pi 格式文件"的路径都不存在。若未来出现只认手工配置文件的 插件(top50 无此形态),标准处置=引擎 config 加 per-package 的 DSH 形状配置槽由中间层翻译落盘——出现第一个消费者时按此补,不预制。
  • Pi 扩展工厂没有参数位ExtensionFactory = (pi) => void, ../pi types.ts 实锤):Pi 官方不存在"装插件给插件传参数"的通道,插件 配置一律由插件自己定义来源(环境变量是事实标准,如 VISION_BRIDGE_*)。 applyPiPackage 的 options.config 只喂中间层自己(visionCompanions), 永远进不了插件视野。别为这个不存在的通道发明 per-package 透传; 若上游 Pi 某天给工厂加了 config 参数,再按 DSH 惯例(管理者插件 config 按名嵌套,llm-pi-ai providers 同款)一步接上。

Read the full file on GitHub · 614 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. 9d ago First seen · 614 lines · 14,464 tokens per session scan C de6a3912b15f

Subscribe to this mod's changes

pi2dsh CLAUDE.md is an instructions file published in the GitHub repository weijiafu14/pi2dsh (185 stars, last pushed 9d ago), licensed MIT. It adds 14,464 tokens to every session, about $0.0723 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other instructions, from other repositories

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,153 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

vscode oss-third-party-notices.instructions.md

Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).

microsoft/vscode · 5,001 tokens

langchain AGENTS.md

AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.

langchain-ai/langchain · 4,469 tokens

spec-kit AGENTS.md

AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.

github/spec-kit · 7,104 tokens