What is kin CLAUDE.md?

A project guide for Kin, a self-hosted workspace where small teams can run Claude-based coding agents privately inside their organisation.

Use it when designing features, choosing software libraries, changing the agent runtime, or reviewing security and deployment decisions.

It records the product's intended users, security assumptions, deployment priorities, and required agent software so new work does not move the project in a different direction.

kin CLAUDE.md — Instructions file for Claude Code by deeptoai-com: A project guide for Kin, a self-hosted workspace where…

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/deeptoai-com/kin/claude-md

Clone the repo

git clone --depth 1 https://github.com/deeptoai-com/kin

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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/deeptoai-com/kin/claude-md.svg)](https://agentmods.dev/instructions/deeptoai-com/kin/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/deeptoai-com/kin/claude-md"><img src="https://agentmods.dev/badge/instructions/deeptoai-com/kin/claude-md.svg" alt="Measured on agentmods" height="20"></a>

Per session 9,414 This file is loaded in full into every session.

When invoked 9,414 The same file — it is already loaded in full.

Security scan C 2 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.

Model	Per session	Once invoked
Fable 5.1	$0.09414	$0.09414
Opus 5	$0.04707	$0.04707
Sonnet 5	$0.01883	$0.01883
Haiku 4.5	$0.00941	$0.00941

Measured 6d ago against content hash 47611ec4af70, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade C, and why

kin CLAUDE.md scanned grade C with 2 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 6d 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.

rm -rf .output dist

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

# 验证：docker inspect Kin-app …Networks 含 kin-private；docker logs Kin-app 无 ENOTFOUND/28P01；curl -sI https://oxygenie.cc → 200

CLAUDE.md · 780 lines

How it starts

The opening of the file, as written. The whole thing — 780 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Kin - 开发规则

历史品牌：本项目早期称为 OxyGenie，仍保留 oxygenie.cc 作为生产域名（Owner 2026-06-12 确认），但代码仓库、GHCR 镜像、README/官方文案均已迁移到 Kin / Deeptoai-com。文中出现 oxygenie.cc / oxygenie-cc2-private 等字样是真实部署事实 或历史坑点，并非待改的"品牌错位"——请勿盲改。

产品定位与设计前提（北极星 · 任何设计/功能先对照此处，避免偏差）

Kin 是面向中小团队的「私有化部署」自治 Claude-Agent 工作台。 典型场景：公司内部 / 团队内部，由可信的小圈子（同事）自托管使用——这也是我们鼓励的用法。它是**自托管、单组织、多用户（组织内多个可信用户）**的产品，不是面向公网匿名大众的多租户 SaaS。

设计自检（每次做设计/功能前对照；若与下列前提冲突，停下来与 Owner 确认）：

威胁模型 = 半可信同事，不是匿名攻击者。 安全是「防误操作 + 共享宿主/密钥卫生 + 组织内用户隔离」的纵深防御，不是反匿名攻击的锁死。触达服务器的强力功能（stdio MCP、连内网/本地工具、代码执行）是合法核心用途——用沙盒 + 警示护栏，而非禁止。
能力是给团队自己用的精选集，不是公开市场。 Skills/MCP 为相对固定的精选，无评分/付费/公开市场（上游 API 只做候选来源，由团队策展）。
必须跑在团队选用的模型/网关上。 默认 ARK（火山）多模型网关；Claude Agent SDK 钉死在 ARK 兼容上限 0.2.112。不要设计依赖「仅原生 Anthropic」的 SDK 0.3.x 特性，除非先有明确的网关/迁移决策。
可部署性优先于超大规模。 为「团队 Docker 一键起跑、开箱即用」优化，而非弹性公网 SaaS 规模。

完整论述见 docs/project/VISION.md §1。此定位为已定（2026-06），变更需明确的战略决策。

项目背景

本项目基于 TanStack Start 构建，提供 SSR、路由、服务端函数等现代全栈能力。

Agent 运行时（单一 SDK）

项目只用 Claude Agent SDK（交互式聊天 + 代码执行 + 真预览，经 WebSocket 持久连接、子进程隔离、Per-Session 沙盒、原生会话恢复）。Mastra 已于 2026-06 彻底移除（连同 playwright / libreoffice，瘦身以恢复免费 CI 构建）——不要再引入第二套 Agent SDK / Vercel AI SDK（ai / @ai-sdk/*），如需新增 LLM 网关能力走「多模型」路线（见 ROADMAP「Later」）。

MCP (Model Context Protocol) 状态

运行时注入已真实可用：resolveMcpServerConfigs（src/claude/mcp/manager.js）在 worker 内把 mcpServers 传给 SDK query()（sdk/stdio/http/sse 四类，见 ws-query-worker.mjs），session metadata 暴露 mcp_servers。不要把下一条误读成"MCP 没做"。
待完善的是策展目录/选择器 UI（团队精选集管理），见 ROADMAP「Next」。

项目状态

本项目基于 TanStack Start 构建。

GitHub 仓库: https://github.com/deeptoai-com/kin

已完成的功能 (Phase 1-4):

Phase 1: WebSocket 服务器 + Claude Agent SDK 集成
Phase 2: 用户隔离（Child Process + Docker 容器化）
Phase 3: Per-Session Sandbox（每会话独立配置目录）
Phase 4: 前端集成（Session 列表、Resume、标题管理）

开发目录规则

正确的开发目录: 项目根目录（仓库 clone 后的目录）

禁止在以下目录开发:

父目录（仅用于文档和参考）
其他项目目录（原始参考代码等）
任何临时开发目录（如历史 phase 目录，已清理）

Read the full file on GitHub · 780 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.

6d ago First seen · 780 lines · 9,414 tokens per session scan C 47611ec4af70

Subscribe to this mod's changes

kin CLAUDE.md is an instructions file published in the GitHub repository deeptoai-com/kin (11 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 9,414 tokens to every session, about $0.0471 per session on Opus 5. A static security scan graded it C with 2 findings (recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

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

kin CLAUDE.md