# adeu-desktop-extension - Doramagic AI Context Pack

> 定位：安装前体验与判断资产。它帮助宿主 AI 有一个好的开始，但不代表已经安装、执行或验证目标项目。

## 充分原则

- **充分原则，不是压缩原则**：AI Context Pack 应该充分到让宿主 AI 在开工前理解项目价值、能力边界、使用入口、风险和证据来源；它可以分层组织，但不以最短摘要为目标。
- **压缩策略**：只压缩噪声和重复内容，不压缩会影响判断和开工质量的上下文。

## 给宿主 AI 的使用方式

你正在读取 Doramagic 为 adeu-desktop-extension 编译的 AI Context Pack。请把它当作开工前上下文：帮助用户理解适合谁、能做什么、如何开始、哪些必须安装后验证、风险在哪里。不要声称你已经安装、运行或执行了目标项目。

## Claim 消费规则

- **事实来源**：Repo Evidence + Claim/Evidence Graph；Human Wiki 只提供显著性、术语和叙事结构。
- **事实最低状态**：`supported`
- `supported`：可以作为项目事实使用，但回答中必须引用 claim_id 和证据路径。
- `weak`：只能作为低置信度线索，必须要求用户继续核实。
- `inferred`：只能用于风险提示或待确认问题，不能包装成项目事实。
- `unverified`：不得作为事实使用，应明确说证据不足。
- `contradicted`：必须展示冲突来源，不得替用户强行选择一个版本。

## 它最适合谁

- **正在使用 Claude/Codex/Cursor/Gemini 等宿主 AI 的开发者**：README 或插件配置提到多个宿主 AI。 证据：`README.md` Claim：`clm_0004` supported 0.86
- **希望把专业流程带进宿主 AI 的用户**：仓库包含 Skill 文档。 证据：`python/skills/adeu-redlining/SKILL.md` Claim：`clm_0005` supported 0.86

## 它能做什么

- **AI Skill / Agent 指令资产库**（可做安装前预览）：项目包含可被宿主 AI 读取的 Skill 或 Agent 指令文件，可用于把专业流程带入 Claude、Codex、Cursor 等宿主。 证据：`python/skills/adeu-redlining/SKILL.md` Claim：`clm_0001` supported 0.86
- **多宿主安装与分发**（需要安装后验证）：项目包含插件或 marketplace 配置，说明它面向一个或多个 AI 宿主的安装和分发。 证据：`.claude-plugin/marketplace.json` Claim：`clm_0002` supported 0.86
- **命令行启动或安装流程**（需要安装后验证）：项目文档中存在可执行命令，真实使用需要在本地或宿主环境中运行这些命令。 证据：`README.md` Claim：`clm_0003` supported 0.86

## 怎么开始

- `npx -y @smithery/cli install adeu --client claude` 证据：`README.md` Claim：`clm_0006` supported 0.86
- `pip install uv` 证据：`README.md` Claim：`clm_0007` supported 0.86
- `pip install langchain-adeu` 证据：`README.md` Claim：`clm_0008` supported 0.86

## 继续前判断卡

- **当前建议**：仅建议沙盒试装
- **为什么**：项目存在安装命令、宿主配置或本地写入线索，不建议直接进入主力环境，应先在隔离环境试装。

### 30 秒判断

- **现在怎么做**：仅建议沙盒试装
- **最小安全下一步**：先跑 Prompt Preview；若仍要安装，只在隔离环境试装
- **先别相信**：真实输出质量不能在安装前相信。
- **继续会触碰**：命令执行、宿主 AI 配置、本地环境或项目文件

### 现在可以相信

- **适合人群线索：正在使用 Claude/Codex/Cursor/Gemini 等宿主 AI 的开发者**（supported）：有 supported claim 或项目证据支撑，但仍不等于真实安装效果。 证据：`README.md` Claim：`clm_0004` supported 0.86
- **适合人群线索：希望把专业流程带进宿主 AI 的用户**（supported）：有 supported claim 或项目证据支撑，但仍不等于真实安装效果。 证据：`python/skills/adeu-redlining/SKILL.md` Claim：`clm_0005` supported 0.86
- **能力存在：AI Skill / Agent 指令资产库**（supported）：可以相信项目包含这类能力线索；是否适合你的具体任务仍要试用或安装后验证。 证据：`python/skills/adeu-redlining/SKILL.md` Claim：`clm_0001` supported 0.86
- **能力存在：多宿主安装与分发**（supported）：可以相信项目包含这类能力线索；是否适合你的具体任务仍要试用或安装后验证。 证据：`.claude-plugin/marketplace.json` Claim：`clm_0002` supported 0.86
- **能力存在：命令行启动或安装流程**（supported）：可以相信项目包含这类能力线索；是否适合你的具体任务仍要试用或安装后验证。 证据：`README.md` Claim：`clm_0003` supported 0.86
- **存在 Quick Start / 安装命令线索**（supported）：可以相信项目文档出现过启动或安装入口；不要因此直接在主力环境运行。 证据：`README.md` Claim：`clm_0006` supported 0.86

### 现在还不能相信

- **真实输出质量不能在安装前相信。**（unverified）：Prompt Preview 只能展示引导方式，不能证明真实项目中的结果质量。
- **宿主 AI 版本兼容性不能在安装前相信。**（unverified）：Claude、Cursor、Codex、Gemini 等宿主加载规则和版本差异必须在真实环境验证。
- **不会污染现有宿主 AI 行为，不能直接相信。**（inferred）：Skill、plugin、AGENTS/CLAUDE/GEMINI 指令可能改变宿主 AI 的默认行为。 证据：`.claude-plugin/marketplace.json`, `GEMINI.md`, `python/skills/adeu-redlining/SKILL.md`
- **可安全回滚不能默认相信。**（unverified）：除非项目明确提供卸载和恢复说明，否则必须先在隔离环境验证。
- **真实安装后是否与用户当前宿主 AI 版本兼容？**（unverified）：兼容性只能通过实际宿主环境验证。 证据：`.claude-plugin/marketplace.json`
- **项目输出质量是否满足用户具体任务？**（unverified）：安装前预览只能展示流程和边界，不能替代真实评测。
- **安装命令是否需要网络、权限或全局写入？**（unverified）：这影响企业环境和个人环境的安装风险。 证据：`README.md`

### 继续会触碰什么

- **命令执行**：包管理器、网络下载、本地插件目录、项目配置或用户主目录。 原因：运行第一条命令就可能产生环境改动；必须先判断是否值得跑。 证据：`README.md`
- **宿主 AI 配置**：Claude/Codex/Cursor/Gemini/OpenCode 等宿主的 plugin、Skill 或规则加载配置。 原因：宿主配置会改变 AI 后续工作方式，可能和用户已有规则冲突。 证据：`.claude-plugin/marketplace.json`, `GEMINI.md`, `python/skills/adeu-redlining/SKILL.md`
- **本地环境或项目文件**：安装结果、插件缓存、项目配置或本地依赖目录。 原因：安装前无法证明写入范围和回滚方式，需要隔离验证。 证据：`.claude-plugin/marketplace.json`, `README.md`
- **宿主 AI 上下文**：AI Context Pack、Prompt Preview、Skill 路由、风险规则和项目事实。 原因：导入上下文会影响宿主 AI 后续判断，必须避免把未验证项包装成事实。

### 最小安全下一步

- **先跑 Prompt Preview**：用安装前交互式试用判断工作方式是否匹配，不需要授权或改环境。（适用：任何项目都适用，尤其是输出质量未知时。）
- **只在隔离目录或测试账号试装**：避免安装命令污染主力宿主 AI、真实项目或用户主目录。（适用：存在命令执行、插件配置或本地写入线索时。）
- **先备份宿主 AI 配置**：Skill、plugin、规则文件可能改变 Claude/Cursor/Codex 的默认行为。（适用：存在插件 manifest、Skill 或宿主规则入口时。）
- **安装后只验证一个最小任务**：先验证加载、兼容、输出质量和回滚，再决定是否深用。（适用：准备从试用进入真实工作流时。）

### 退出方式

- **保留安装前状态**：记录原始宿主配置和项目状态，后续才能判断是否可恢复。
- **准备移除宿主 plugin / Skill / 规则入口**：如果试装后行为异常，可以把宿主 AI 恢复到试装前状态。
- **记录安装命令和写入路径**：没有明确卸载说明时，至少要知道哪些目录或配置需要手动清理。
- **如果没有回滚路径，不进入主力环境**：不可回滚是继续前阻断项，不应靠信任或运气继续。

## 哪些只能预览

- 解释项目适合谁和能做什么
- 基于项目文档演示典型对话流程
- 帮助用户判断是否值得安装或继续研究

## 哪些必须安装后验证

- 真实安装 Skill、插件或 CLI
- 执行脚本、修改本地文件或访问外部服务
- 验证真实输出质量、性能和兼容性

## 边界与风险判断卡

- **把安装前预览误认为真实运行**：用户可能高估项目已经完成的配置、权限和兼容性验证。 处理方式：明确区分 prompt_preview_can_do 与 runtime_required。 Claim：`clm_0009` inferred 0.45
- **宿主 AI 插件或 Skill 规则冲突**：新规则可能改变用户现有宿主 AI 的工作方式。 处理方式：安装前先检查插件 manifest 和 Skill 文件，必要时隔离测试。 证据：`.claude-plugin/marketplace.json` Claim：`clm_0010` supported 0.86
- **命令执行会修改本地环境**：安装命令可能写入用户主目录、宿主插件目录或项目配置。 处理方式：先在隔离环境或测试账号中运行。 证据：`README.md` Claim：`clm_0011` supported 0.86
- **待确认**：真实安装后是否与用户当前宿主 AI 版本兼容？。原因：兼容性只能通过实际宿主环境验证。
- **待确认**：项目输出质量是否满足用户具体任务？。原因：安装前预览只能展示流程和边界，不能替代真实评测。
- **待确认**：安装命令是否需要网络、权限或全局写入？。原因：这影响企业环境和个人环境的安装风险。

## 开工前工作上下文

### 加载顺序

- 先读取 how_to_use.host_ai_instruction，建立安装前判断资产的边界。
- 读取 claim_graph_summary，确认事实来自 Claim/Evidence Graph，而不是 Human Wiki 叙事。
- 再读取 intended_users、capabilities 和 quick_start_candidates，判断用户是否匹配。
- 需要执行具体任务时，优先查 role_skill_index，再查 evidence_index。
- 遇到真实安装、文件修改、网络访问、性能或兼容性问题时，转入 risk_card 和 boundaries.runtime_required。

### 任务路由

- **AI Skill / Agent 指令资产库**：先基于 role_skill_index / evidence_index 帮用户挑选可用角色、Skill 或工作流。 边界：可做安装前 Prompt 体验。 证据：`python/skills/adeu-redlining/SKILL.md` Claim：`clm_0001` supported 0.86
- **多宿主安装与分发**：先说明这是安装后验证能力，再给出安装前检查清单。 边界：必须真实安装或运行后验证。 证据：`.claude-plugin/marketplace.json` Claim：`clm_0002` supported 0.86
- **命令行启动或安装流程**：先说明这是安装后验证能力，再给出安装前检查清单。 边界：必须真实安装或运行后验证。 证据：`README.md` Claim：`clm_0003` supported 0.86

### 上下文规模

- 文件总数：258
- 重要文件覆盖：40/258
- 证据索引条目：54
- 角色 / Skill 条目：1

### 证据不足时的处理

- **missing_evidence**：说明证据不足，要求用户提供目标文件、README 段落或安装后验证记录；不要补全事实。
- **out_of_scope_request**：说明该任务超出当前 AI Context Pack 证据范围，并建议用户先查看 Human Manual 或真实安装后验证。
- **runtime_request**：给出安装前检查清单和命令来源，但不要替用户执行命令或声称已执行。
- **source_conflict**：同时展示冲突来源，标记为待核实，不要强行选择一个版本。

## Prompt Recipes

### 适配判断

- 目标：判断这个项目是否适合用户当前任务。
- 预期输出：适配结论、关键理由、证据引用、安装前可预览内容、必须安装后验证内容、下一步建议。

```text
请基于 adeu-desktop-extension 的 AI Context Pack，先问我 3 个必要问题，然后判断它是否适合我的任务。回答必须包含：适合谁、能做什么、不能做什么、是否值得安装、证据来自哪里。所有项目事实必须引用 evidence_refs、source_paths 或 claim_id。
```

### 安装前体验

- 目标：让用户在安装前感受核心工作流，同时避免把预览包装成真实能力或营销承诺。
- 预期输出：一段带边界标签的体验剧本、安装后验证清单和谨慎建议；不含真实运行承诺或强营销表述。

```text
请把 adeu-desktop-extension 当作安装前体验资产，而不是已安装工具或真实运行环境。

请严格输出四段：
1. 先问我 3 个必要问题。
2. 给出一段“体验剧本”：用 [安装前可预览]、[必须安装后验证]、[证据不足] 三种标签展示它可能如何引导工作流。
3. 给出安装后验证清单：列出哪些能力只有真实安装、真实宿主加载、真实项目运行后才能确认。
4. 给出谨慎建议：只能说“值得继续研究/试装”“先补充信息后再判断”或“不建议继续”，不得替项目背书。

硬性边界：
- 不要声称已经安装、运行、执行测试、修改文件或产生真实结果。
- 不要写“自动适配”“确保通过”“完美适配”“强烈建议安装”等承诺性表达。
- 如果描述安装后的工作方式，必须使用“如果安装成功且宿主正确加载 Skill，它可能会……”这种条件句。
- 体验剧本只能写成“示例台词/假设流程”：使用“可能会询问/可能会建议/可能会展示”，不要写“已写入、已生成、已通过、正在运行、正在生成”。
- Prompt Preview 不负责给安装命令；如用户准备试装，只能提示先阅读 Quick Start 和 Risk Card，并在隔离环境验证。
- 所有项目事实必须来自 supported claim、evidence_refs 或 source_paths；inferred/unverified 只能作风险或待确认项。

```

### 角色 / Skill 选择

- 目标：从项目里的角色或 Skill 中挑选最匹配的资产。
- 预期输出：候选角色或 Skill 列表，每项包含适用场景、证据路径、风险边界和是否需要安装后验证。

```text
请读取 role_skill_index，根据我的目标任务推荐 3-5 个最相关的角色或 Skill。每个推荐都要说明适用场景、可能输出、风险边界和 evidence_refs。
```

### 风险预检

- 目标：安装或引入前识别环境、权限、规则冲突和质量风险。
- 预期输出：环境、权限、依赖、许可、宿主冲突、质量风险和未知项的检查清单。

```text
请基于 risk_card、boundaries 和 quick_start_candidates，给我一份安装前风险预检清单。不要替我执行命令，只说明我应该检查什么、为什么检查、失败会有什么影响。
```

### 宿主 AI 开工指令

- 目标：把项目上下文转成一次对话开始前的宿主 AI 指令。
- 预期输出：一段边界明确、证据引用明确、适合复制给宿主 AI 的开工前指令。

```text
请基于 adeu-desktop-extension 的 AI Context Pack，生成一段我可以粘贴给宿主 AI 的开工前指令。这段指令必须遵守 not_runtime=true，不能声称项目已经安装、运行或产生真实结果。
```


## 角色 / Skill 索引

- 共索引 1 个角色 / Skill / 项目文档条目。

- **adeu-redlining**（skill）：Use this skill for reviewing, editing, or negotiating existing Word documents .docx where "Track Changes" or precise redlining is required. Use it to propose edits, accept/reject changes, or reply to comments. Do NOT use for creating new blank documents from scratch use docx skill for that . 激活提示：当用户任务与“adeu-redlining”描述的流程高度相关时，先用它做安装前体验，再决定是否安装。 证据：`python/skills/adeu-redlining/SKILL.md`

## 证据索引

- 共索引 54 条证据。

- **Adeu — Native Track Changes for AI**（documentation）：Adeu is your document redlining engine. It gives you a safe, token-efficient interface to read and edit .docx files, producing native Microsoft Word Track Changes rather than rewriting the file. 证据：`GEMINI.md`
- **Adeu: Track Changes for the LLM era**（documentation）：Adeu: Track Changes for the LLM era 证据：`README.md`
- **Adeu Ecosystem & Integrations**（documentation）：Welcome to the Adeu Ecosystem directory. 证据：`ecosystem/README.md`
- **langchain-adeu**（documentation）：! PyPI version https://img.shields.io/pypi/v/langchain-adeu.svg https://pypi.org/project/langchain-adeu/ ! License: MIT https://img.shields.io/badge/License-MIT-blue.svg https://opensource.org/licenses/MIT 证据：`langchain/README.md`
- **@adeu/core**（documentation）：! GitHub Repo stars https://img.shields.io/github/stars/dealfluence/adeu?style=social https://github.com/dealfluence/adeu ! License: MIT https://img.shields.io/badge/License-MIT-blue.svg https://opensource.org/licenses/MIT 证据：`node/packages/core/README.md`
- **@adeu/mcp-server**（documentation）：! GitHub Repo stars https://img.shields.io/github/stars/dealfluence/adeu?style=social https://github.com/dealfluence/adeu ! MCP Compatible https://img.shields.io/badge/MCP-Compatible-green.svg https://modelcontextprotocol.io/ ! License: MIT https://img.shields.io/badge/License-MIT-blue.svg https://opensource.org/licenses/MIT 证据：`node/packages/mcp-server/README.md`
- **n8n-nodes-adeu**（documentation）：! npm version https://img.shields.io/npm/v/n8n-nodes-adeu.svg https://www.npmjs.com/package/n8n-nodes-adeu ! License: MIT https://img.shields.io/badge/License-MIT-blue.svg https://opensource.org/licenses/MIT 证据：`node/packages/n8n-nodes-adeu/README.md`
- **Package**（package_manifest）：{ "name": "adeu-desktop-extension", "type": "module", "private": true, "engines": { "node": " =22.0.0" } } 证据：`desktop-extension/package.json`
- **Package**（package_manifest）：{ "name": "adeu-workspace", "private": true, "workspaces": "packages/ " , "scripts": { "build": "npm run build --workspaces --if-present", "test": "npm run test --workspaces --if-present" }, "engines": { "node": " =22.0.0" }, "devDependencies": { "@types/node": "^25.6.2", "tsup": "^8.5.1", "typescript": "^6.0.3", "vitest": "^4.1.5" } } 证据：`node/package.json`
- **Contributing to Adeu**（documentation）：Thank you for your interest in contributing to Adeu! We welcome bug reports, feature requests, and pull requests from the community. 证据：`CONTRIBUTING.md`
- **Package**（package_manifest）：{ "name": "@adeu/core", "version": "1.7.5", "description": "", "main": "./dist/index.js", "types": "./dist/index.d.ts", "exports": { ".": { "types": "./dist/index.d.ts", "import": "./dist/index.js", "require": "./dist/index.cjs" } }, "scripts": { "build": "tsup", "test": "vitest run" }, "repository": { "type": "git", "url": "https://github.com/dealfluence/adeu.git", "directory": "node/packages/core" }, "engines": { "node": " =22.0.0" }, "publishConfig": { "access": "public" }, "author": "Mikko Korpela ", "license": "MIT", "type": "module", "dependencies": { "@xmldom/xmldom": "^0.9.10", "diff-match-patch": "^1.0.5", "fflate": "^0.8.2", "xpath": "^0.0.34" }, "devDependencies": { "@types/diff-… 证据：`node/packages/core/package.json`
- **Package**（package_manifest）：{ "name": "@adeu/mcp-server", "version": "1.7.5", "description": "", "mcpName": "ai.adeu/adeu", "main": "./dist/index.js", "types": "./dist/index.d.ts", "bin": { "adeu-mcp-server": "./dist/index.js" }, "devDependencies": {}, "scripts": { "build": "tsup", "test": "vitest run", "start": "node ./dist/index.js" }, "repository": { "type": "git", "url": "https://github.com/dealfluence/adeu.git", "directory": "node/packages/mcp-server" }, "engines": { "node": " =22.0.0" }, "publishConfig": { "access": "public" }, "author": "Mikko Korpela ", "license": "MIT", "type": "module", "dependencies": { "@adeu/core": "^1.0.0", "@modelcontextprotocol/sdk": "^1.0.0" } } 证据：`node/packages/mcp-server/package.json`
- **Package**（package_manifest）：{ "name": "n8n-nodes-adeu", "version": "1.7.5", "description": "n8n community node for Adeu — apply AI-driven tracked changes, generate diffs, and sanitize Microsoft Word .docx documents.", "keywords": "n8n-community-node-package", "n8n", "adeu", "docx", "word", "redline", "track-changes", "diff", "openxml" , "license": "MIT", "homepage": "https://adeu.ai", "author": { "name": "Mikko Korpela", "email": "mikko@adeu.ai" }, "repository": { "type": "git", "url": "git+https://github.com/dealfluence/adeu.git", "directory": "node/packages/n8n-nodes-adeu" }, "engines": { "node": " =22.0.0" }, "main": "dist/nodes/Adeu/Adeu.node.js", "scripts": { "build": "tsup && npm run copy-assets", "copy-assets":… 证据：`node/packages/n8n-nodes-adeu/package.json`
- **Adeu: Professional Redlining & Review**（skill_instruction）：Adeu: Professional Redlining & Review 证据：`python/skills/adeu-redlining/SKILL.md`
- **Marketplace**（structured_config）：{ "name": "adeu-skills", "owner": { "name": "Adeu Solutions", "url": "https://github.com/dealfluence/adeu" }, "metadata": { "description": "Agentic tools for reviewing, redlining, and negotiating DOCX contracts.", "version": "0.10.0" }, "plugins": { "name": "adeu-redlining", "description": "Professional DOCX Redlining: Track Changes, Comments, and Review Logic.", "source": "./", "strict": false, "skills": "./skills/adeu-redlining" } } 证据：`.claude-plugin/marketplace.json`
- **License**（source_file）：Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files the "Software" , to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: 证据：`LICENSE`
- **Architecture: Document Reconciliation Virtual DOM**（documentation）：Architecture: Document Reconciliation Virtual DOM 证据：`docs/ARCHITECTURE.md`
- **Free Agent Problem**（documentation）：The Free Agent Leaky Abstraction: Architectural and Cognitive Containment of LLMsIntroduction to the Leaky Abstraction in Agentic AIThe rapid evolution of artificial intelligence from static, text-generating foundation models to autonomous, tool-wielding agentic systems has introduced a profound paradigm shift in computational architecture. Large Language Models LLMs integrated with execution environments and tool-calling capabilities through standardizations such as the Model Context Protocol MCP are increasingly deployed to autonomously orchestrate workflows, query databases, manipulate file systems, and interface with third-party application programming interfaces APIs . However, grantin… 证据：`docs/FREE_AGENT_PROBLEM.md`
- **Model Context Protocol MCP Apps Extension**（documentation）：Model Context Protocol MCP Apps Extension 证据：`docs/MCP_APPS_PROTOCOL.md`
- **Adeu MCP Server Authentication Architecture**（documentation）：Here is a detailed, concise Markdown document outlining the authentication architecture we just built. You can save this as docs/MCP AUTH ARCHITECTURE.md or add it to your ARCHITECTURE.md . 证据：`docs/MCP_AUTH_ARCHITECTURE.md`
- **Adeu Publishing & Registry Guidelines**（documentation）：Adeu Publishing & Registry Guidelines 证据：`docs/PUBLISHING.md`
- **SeverityAreaTitle1🔴 Majorread docx outlineOutline detector misses long custom-derived heading styles StyleHeading2NotIt…**（documentation）：QA Findings Summary I'll group by severity. All findings include the expectation I had before each call; full repros are in the chat above. SeverityAreaTitle1🔴 Majorread docx outlineOutline detector misses long custom-derived heading styles StyleHeading2NotItalicBefore0ptAfter0ptLinespa not recognized; the shorter StyleHeading2NotItalic is . example03's "Sub Headings:" H2 is missing from outline.6🔴 Major perf process document batch live modeSingle live edit took 50.034s vs 0.055s on disk ~900× slower . Other live calls finished in heading marker on one line, text on next . Breaks target text=" Foo" matches in subsequent edits.14🟡 Minorlive WordLive-mode latency varies wildly 0.4s — 50s — sa… 证据：`docs/QA_ISSUES_DISCOVERED.md`
- **FILE: TODO.md**（documentation）：1. Document Scope Expansion Complex Fields : Investigate safe strategies for handling Fields Date, TOC without breaking them. 证据：`docs/TODO.md`
- **Architectural Impacts of Anthropic's claude-for-legal on Adeu**（documentation）：Architectural Impacts of Anthropic's claude-for-legal on Adeu 证据：`docs/claude-for-legal-impact-analysis.md`
- **Specification: Polyglot Consistency Framework**（documentation）：Specification: Polyglot Consistency Framework 证据：`docs/spec-consistency-framework.md`
- **Feature Specification: Document Finalization**（documentation）：Feature Specification: Document Finalization 证据：`docs/spec-finalization.md`
- **Specification: Polyglot Monorepo & Native Node.js Port**（documentation）：Specification: Polyglot Monorepo & Native Node.js Port 证据：`docs/spec-monorepo-node-port.md`
- **Specification: Structural Table Edits**（documentation）：Specification: Structural Table Edits 证据：`docs/spec-table-edits.md`
- **FILE: adeu/docs/spec-undo-record.md**（documentation）：FILE: adeu/docs/spec-undo-record.md Specification: Live Word COM UndoRecord Integration 证据：`docs/spec-undo-record.md`
- **Project Adeu: Technical Specification**（documentation）：Project Adeu: Technical Specification 证据：`docs/spec.md`
- **Pull Request Overview**（documentation）：Please provide a brief description of the changes introduced by this PR. 证据：`.github/PULL_REQUEST_TEMPLATE.md`
- **Project Context: Adeu**（documentation）：System Overview Adeu acts as a "Virtual DOM" for DOCX files, enabling LLMs to edit documents via a text proxy while preserving complex XML structure. - Ingestion : ingest.py creates a Markdown/CriticMarkup representation of the document. - Mapping : mapper.py builds a linear index of text spans linking back to python-docx objects. - Reconciliation : engine.py calculates and applies atomic XML patches w:ins / w:del . - Agent Interface : server.py exposes these capabilities as an MCP Model Context Protocol server, while cli.py handles automated environment configuration. 证据：`AI_CONTEXT.md`
- **Adeu Ecosystem: Vendor & Integration Policy**（documentation）：Adeu Ecosystem: Vendor & Integration Policy 证据：`ecosystem/VENDOR_POLICY.md`
- **Golden Clean**（documentation）：This is the final contract 证据：`shared/cross_platform_tests/01_basic_text_modification/golden_clean.md`
- **Golden Raw**（documentation）：This is the {--initial document--}{++final contract++}{ Chg:1 delete Adeu AI Chg:2 insert Adeu AI Com:1 Adeu AI @ DATE: Updated terminology.<<} 证据：`shared/cross_platform_tests/01_basic_text_modification/golden_raw.md`
- **Settings**（structured_config）：{ "mcpServers": { "adeu": { "command": "uv", "args": "run", "adeu-server" , "env": { "ADEU ENABLE TEST TOOLS": "1" } } } } 证据：`.gemini/settings.json`
- **Manifest**（structured_config）：{ "manifest version": "0.2", "name": "Adeu", "version": "1.7.5", "description": "Native Node.js DOCX LLM Context Engine No Python Required .", "author": { "name": "mikko@adeu.ai", "email": "mikko@adeu.ai", "url": "https://adeu.ai" }, "homepage": "https://adeu.ai", "icon": "icon.png", "server": { "type": "node", "entry point": "index.js", "mcp config": { "command": "node", "args": "${ dirname}/index.js" , "env": {} } }, "license": "MIT", "repository": { "type": "git", "url": "https://github.com/dealfluence/adeu" } } 证据：`desktop-extension/manifest.json`
- **Gemini Extension**（structured_config）：{ "name": "adeu", "version": "1.7.5", "contextFileName": "GEMINI.md", "mcpServers": { "adeu": { "command": "node", "args": "${extensionPath}${/}index.js" , "cwd": "${extensionPath}" } } } 证据：`gemini-extension.json`
- **Server**（structured_config）：{ "$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "name": "ai.adeu/adeu", "description": "Automated DOCX Redlining Engine", "repository": { "url": "https://github.com/dealfluence/adeu", "source": "github" }, "version": "1.7.5", "packages": { "registryType": "npm", "identifier": "@adeu/mcp-server", "version": "1.7.5", "transport": { "type": "stdio" } } } 证据：`python/server.json`
- **Tsconfig**（structured_config）：{ "compilerOptions": { "target": "ES2022", "module": "NodeNext", "moduleResolution": "NodeNext", "lib": "ES2022", "DOM" , "types": "node" , "ignoreDeprecations": "6.0", "rootDir": "./src", "outDir": "./dist", "declaration": true, "declarationMap": true, "sourceMap": true, "strict": true, "esModuleInterop": true, "skipLibCheck": true, "forceConsistentCasingInFileNames": true }, "include": "src/ / " , "exclude": "node modules", "dist" } 证据：`node/packages/core/tsconfig.json`
- **Tsconfig**（structured_config）：{ "compilerOptions": { "target": "ES2022", "module": "NodeNext", "moduleResolution": "NodeNext", "lib": "ES2022", "DOM" , "types": "node" , "ignoreDeprecations": "6.0", "rootDir": "./src", "outDir": "./dist", "declaration": true, "declarationMap": true, "sourceMap": true, "strict": true, "esModuleInterop": true, "skipLibCheck": true, "forceConsistentCasingInFileNames": true }, "include": "src/ / " , "exclude": "node modules", "dist" } 证据：`node/packages/mcp-server/tsconfig.json`
- **Adeu.Node**（structured_config）：{ "node": "n8n-nodes-adeu.adeu", "nodeVersion": "1.0", "codexVersion": "1.0", "categories": "Productivity", "Utility" , "resources": { "primaryDocumentation": { "url": "https://github.com/dealfluence/adeu readme" } } } 证据：`node/packages/n8n-nodes-adeu/nodes/Adeu/Adeu.node.json`
- **Tsconfig**（structured_config）：{ "compilerOptions": { "target": "ES2022", "module": "CommonJS", "moduleResolution": "Node", "lib": "ES2022", "DOM" , "types": "node" , "ignoreDeprecations": "6.0", "rootDir": "./nodes", "outDir": "./dist/nodes", "declaration": true, "sourceMap": true, "strict": true, "esModuleInterop": true, "skipLibCheck": true, "forceConsistentCasingInFileNames": true, "resolveJsonModule": true }, "include": "nodes/ / " , "exclude": "node modules", "dist", "test", " / .test.ts" } 证据：`node/packages/n8n-nodes-adeu/tsconfig.json`
- **Maintain Python dependencies Native support for pyproject.toml & uv.lock**（source_file）：version: 2 updates: Maintain Python dependencies Native support for pyproject.toml & uv.lock - package-ecosystem: "pip" directory: "/" schedule: interval: "weekly" 证据：`.github/dependabot.yml`
- **Byte-compiled / optimized / DLL files**（source_file）：Byte-compiled / optimized / DLL files pycache / .py codz $py.class 证据：`.gitignore`
- **FILE: langchain/.gitignore**（source_file）：FILE: langchain/.gitignore Python build artifacts dist/ build/ .egg-info/ pycache / .py cod $py.class 证据：`langchain/.gitignore`
- **Makefile**（source_file）：.PHONY: help install test integration test lint format typecheck clean 证据：`langchain/Makefile`
- **Local development: use the sibling adeu package directly so changes**（source_file）：project name = "langchain-adeu" version = "1.0.0" description = "LangChain integration for Adeu — track-changes for DOCX in the LLM era" authors = { name = "Mikko Korpela" }, { name = "Uzair Ahmed" } readme = "README.md" requires-python = " =3.12" license = { file = "../LICENSE" } keywords = "langchain", "docx", "redlining", "tracked-changes", "agent", "tools", "word" classifiers = "Development Status :: 5 - Production/Stable", "Intended Audience :: Developers", "License :: OSI Approved :: MIT License", "Programming Language :: Python :: 3.12", "Programming Language :: Python :: 3.13", "Operating System :: OS Independent", "Topic :: Office/Business :: Office Suites", "Topic :: Scientific/En… 证据：`langchain/pyproject.toml`
- **FILE: .pre-commit-config.yaml**（source_file）：FILE: .pre-commit-config.yaml repos: - repo: local hooks: - id: ruff name: ruff entry: uv run ruff check --force-exclude --fix language: system types or: python, pyi require serial: true 证据：`python/.pre-commit-config.yaml`
- **Copy the project files**（source_file）：FROM ghcr.io/astral-sh/uv:python3.12-bookworm-slim 证据：`python/Dockerfile`
- **Pyproject**（source_file）：project name = "adeu" version = "1.7.5" description = "Automated DOCX Redlining Engine" authors = { name = "Mikko Korpela" }, { name = "Uzair Ahmed" } readme = "../README.md" requires-python = " =3.12" license = { file = "../LICENSE" } keywords = "docx", "redlining", "llm", "mcp", "agent", "virtual-dom" classifiers = "Development Status :: 5 - Production/Stable", "Intended Audience :: Developers", "License :: OSI Approved :: MIT License", "Programming Language :: Python :: 3.12", "Programming Language :: Python :: 3.13", "Operating System :: OS Independent", dependencies = "python-docx =1.1.0", "structlog =24.0.0", "pydantic =2.0.0", "lxml =5.0.0", "diff-match-patch =20230430", "keyring =25… 证据：`python/pyproject.toml`
- **Regex replace to preserve exact file formatting indents/newlines**（source_file）：import json import re import subprocess import sys from pathlib import Path 证据：`scripts/bump.py`
- **We capture stderr so we can actually see if Node crashes**（source_file）：import json import os import subprocess import sys import traceback import zipfile 证据：`scripts/patch_smithery_mcpb.py`
- **Ensure we are running from the repository root**（source_file）：Ensure we are running from the repository root if -not Test-Path "python\server.json" { Write-Host " ERROR Please run this script from the repository root: .\scripts\publish mcp registry.ps1" -ForegroundColor Red exit 1 } 证据：`scripts/publish_mcp_registry.ps1`

## 宿主 AI 必须遵守的规则

- **把本资产当作开工前上下文，而不是运行环境。**：AI Context Pack 只包含证据化项目理解，不包含目标项目的可执行状态。 证据：`GEMINI.md`, `README.md`, `ecosystem/README.md`
- **回答用户时区分可预览内容与必须安装后才能验证的内容。**：安装前体验的消费者价值来自降低误装和误判，而不是伪装成真实运行。 证据：`GEMINI.md`, `README.md`, `ecosystem/README.md`

## 用户开工前应该回答的问题

- 你准备在哪个宿主 AI 或本地环境中使用它？
- 你只是想先体验工作流，还是准备真实安装？
- 你最在意的是安装成本、输出质量、还是和现有规则的冲突？

## 验收标准

- 所有能力声明都能回指到 evidence_refs 中的文件路径。
- AI_CONTEXT_PACK.md 没有把预览包装成真实运行。
- 用户能在 3 分钟内看懂适合谁、能做什么、如何开始和风险边界。

---

## Doramagic Context Augmentation

The following material strengthens the Repomix/AI Context Pack body. Human Manual is only a reading skeleton; pitfall logs become hard operating constraints for the host AI.

## Human Manual Skeleton

Usage rule: this is only a reading path and salience signal, not factual authority. Concrete facts must still come from repo evidence / Claim Graph.

Hard rules for the host AI:
- Do not treat page titles, order, summaries, or importance as project facts.
- When explaining the Human Manual skeleton, state that it is only a reading path / salience signal.
- Capability, installation, compatibility, runtime status, and risk judgments must cite repo evidence, source paths, or Claim Graph.

- **Repository Overview**：importance `high`
  - source_paths: README.md, desktop-extension/package.json, ecosystem/README.md, langchain/README.md, langchain/pyproject.toml
- **Entrypoints and Runtime Surface**：importance `high`
  - source_paths: README.md, desktop-extension/package.json, ecosystem/README.md, langchain/README.md, langchain/pyproject.toml
- **Architecture Evidence Map**：importance `high`
  - source_paths: README.md, desktop-extension/package.json, ecosystem/README.md, langchain/README.md, langchain/pyproject.toml
- **Operations and Verification Boundaries**：importance `high`
  - source_paths: README.md, desktop-extension/package.json, ecosystem/README.md, langchain/README.md, langchain/pyproject.toml

## Repo Inspection Evidence

- repo_clone_verified: true
- repo_inspection_verified: true
- repo_commit: `ac7d4e195bf26a4c46594e5a6c12f34eba9c1e42`
- inspected_files: `README.md`, `docs/MCP_APPS_PROTOCOL.md`, `docs/claude-for-legal-impact-analysis.md`, `docs/FREE_AGENT_PROBLEM.md`, `docs/QA_ISSUES_DISCOVERED.md`, `docs/ARCHITECTURE.md`, `docs/spec.md`, `docs/MCP_AUTH_ARCHITECTURE.md`, `docs/spec-finalization.md`, `docs/spec-undo-record.md`, `docs/spec-table-edits.md`, `docs/spec-monorepo-node-port.md`, `docs/PUBLISHING.md`, `docs/spec-consistency-framework.md`, `docs/TODO.md`, `src/adeu/redline/engine.py`

Hard rules for the host AI:
- Without repo_clone_verified=true, do not claim the source code has been read.
- Without repo_inspection_verified=true, do not turn README/docs/package observations into facts.
- Without quick_start_verified=true, do not claim the Quick Start has been successfully run.

## Doramagic Pitfall Constraints

These rules come from Doramagic discovery, validation, or compilation pitfalls. The host AI must treat them as operating constraints, not general background notes.

### Constraint 1: 来源证据：Optional clause-validation hook for jurisdiction-aware redlines

- Trigger: GitHub 社区证据显示该项目存在一个安装相关的待验证问题：Optional clause-validation hook for jurisdiction-aware redlines
- Host AI rule: 来源问题仍为 open，Pack Agent 需要复核是否仍影响当前版本。
- Why it matters: 可能增加新用户试用和生产接入成本。
- Evidence: community_evidence:github | cevd_d958d22cf43e498a9e36cecc04155467 | https://github.com/dealfluence/adeu/issues/20 | 来源类型 github_issue 暴露的待验证使用条件。
- Hard boundary: do not present this pitfall as solved, verified, or safe to ignore unless later validation evidence explicitly closes it.

### Constraint 2: 仓库名和安装名不一致

- Trigger: 仓库名 `adeu` 与安装入口 `@adeu/mcp-server` 不完全一致。
- Host AI rule: 在 npm/PyPI/GitHub 上确认包名映射和官方 README 说明。
- Why it matters: 用户照着仓库名搜索包或照着包名找仓库时容易走错入口。
- Evidence: identity.distribution | mcp_registry:ai.adeu/adeu:1.7.1 | https://registry.modelcontextprotocol.io/v0.1/servers/ai.adeu%2Fadeu/versions/1.7.1 | repo=adeu; install=@adeu/mcp-server
- Hard boundary: do not present this pitfall as solved, verified, or safe to ignore unless later validation evidence explicitly closes it.

### Constraint 3: 来源证据：Create a non-AGPL workflow for VS Code visualization

- Trigger: GitHub 社区证据显示该项目存在一个配置相关的待验证问题：Create a non-AGPL workflow for VS Code visualization
- Host AI rule: 来源显示可能已有修复、规避或版本变化，说明书中必须标注适用版本。
- Why it matters: 可能增加新用户试用和生产接入成本。
- Evidence: community_evidence:github | cevd_f1676fd51c1a4448920e55baae10f045 | https://github.com/dealfluence/adeu/issues/1 | 来源讨论提到 python 相关条件，需在安装/试用前复核。
- Hard boundary: do not present this pitfall as solved, verified, or safe to ignore unless later validation evidence explicitly closes it.

### Constraint 4: 能力判断依赖假设

- Trigger: README/documentation is current enough for a first validation pass.
- Host AI rule: 将假设转成下游验证清单。
- Why it matters: 假设不成立时，用户拿不到承诺的能力。
- Evidence: capability.assumptions | mcp_registry:ai.adeu/adeu:1.7.1 | https://registry.modelcontextprotocol.io/v0.1/servers/ai.adeu%2Fadeu/versions/1.7.1 | README/documentation is current enough for a first validation pass.
- Hard boundary: do not present this pitfall as solved, verified, or safe to ignore unless later validation evidence explicitly closes it.

### Constraint 5: 维护活跃度未知

- Trigger: 未记录 last_activity_observed。
- Host AI rule: 补 GitHub 最近 commit、release、issue/PR 响应信号。
- Why it matters: 新项目、停更项目和活跃项目会被混在一起，推荐信任度下降。
- Evidence: evidence.maintainer_signals | mcp_registry:ai.adeu/adeu:1.7.1 | https://registry.modelcontextprotocol.io/v0.1/servers/ai.adeu%2Fadeu/versions/1.7.1 | last_activity_observed missing
- Hard boundary: do not present this pitfall as solved, verified, or safe to ignore unless later validation evidence explicitly closes it.

### Constraint 6: 下游验证发现风险项

- Trigger: no_demo
- Host AI rule: 进入安全/权限治理复核队列。
- Why it matters: 下游已经要求复核，不能在页面中弱化。
- Evidence: downstream_validation.risk_items | mcp_registry:ai.adeu/adeu:1.7.1 | https://registry.modelcontextprotocol.io/v0.1/servers/ai.adeu%2Fadeu/versions/1.7.1 | no_demo; severity=medium
- Hard boundary: do not present this pitfall as solved, verified, or safe to ignore unless later validation evidence explicitly closes it.

### Constraint 7: 存在评分风险

- Trigger: no_demo
- Host AI rule: 把风险写入边界卡，并确认是否需要人工复核。
- Why it matters: 风险会影响是否适合普通用户安装。
- Evidence: risks.scoring_risks | mcp_registry:ai.adeu/adeu:1.7.1 | https://registry.modelcontextprotocol.io/v0.1/servers/ai.adeu%2Fadeu/versions/1.7.1 | no_demo; severity=medium
- Hard boundary: do not present this pitfall as solved, verified, or safe to ignore unless later validation evidence explicitly closes it.

### Constraint 8: 来源证据：Claude Opus 4.7 testers feedback

- Trigger: GitHub 社区证据显示该项目存在一个安全/权限相关的待验证问题：Claude Opus 4.7 testers feedback
- Host AI rule: 来源显示可能已有修复、规避或版本变化，说明书中必须标注适用版本。
- Why it matters: 可能影响授权、密钥配置或安全边界。
- Evidence: community_evidence:github | cevd_6508f8a8b81b40c5890e02ea1110c791 | https://github.com/dealfluence/adeu/issues/11 | 来源类型 github_issue 暴露的待验证使用条件。
- Hard boundary: do not present this pitfall as solved, verified, or safe to ignore unless later validation evidence explicitly closes it.

### Constraint 9: 来源证据：Multi-Round Negotiation Testing

- Trigger: GitHub 社区证据显示该项目存在一个安全/权限相关的待验证问题：Multi-Round Negotiation Testing
- Host AI rule: 来源显示可能已有修复、规避或版本变化，说明书中必须标注适用版本。
- Why it matters: 可能影响授权、密钥配置或安全边界。
- Evidence: community_evidence:github | cevd_ce563606cbda4f2d9273aba3173a2ecc | https://github.com/dealfluence/adeu/issues/4 | 来源类型 github_issue 暴露的待验证使用条件。
- Hard boundary: do not present this pitfall as solved, verified, or safe to ignore unless later validation evidence explicitly closes it.

### Constraint 10: 来源证据：The Multiple Chg IDs Per Logical Edit

- Trigger: GitHub 社区证据显示该项目存在一个安全/权限相关的待验证问题：The Multiple Chg IDs Per Logical Edit
- Host AI rule: 来源显示可能已有修复、规避或版本变化，说明书中必须标注适用版本。
- Why it matters: 可能影响授权、密钥配置或安全边界。
- Evidence: community_evidence:github | cevd_8bcb3776b02a46479a29535f3048d7f7 | https://github.com/dealfluence/adeu/issues/16 | 来源类型 github_issue 暴露的待验证使用条件。
- Hard boundary: do not present this pitfall as solved, verified, or safe to ignore unless later validation evidence explicitly closes it.