# visionaire-engine - 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 visionaire-engine 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_0002` supported 0.86

## 它能做什么

- **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_0001` supported 0.86

## 怎么开始

- `claude mcp add visionaire -- npx -y visionaire-engine` 证据：`README.md` Claim：`clm_0003` supported 0.86
- `git clone https://github.com/mi60dev/visionaire-engine && cd visionaire-engine` 证据：`README.md` Claim：`clm_0004` supported 0.86
- `npm install && npm run build` 证据：`README.md` Claim：`clm_0005` supported 0.86
- `claude mcp add visionaire -- node "$PWD/dist/index.js"` 证据：`README.md` Claim：`clm_0006` supported 0.86

## 继续前判断卡

- **当前建议**：先做权限沙盒试用
- **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**：先做权限沙盒试用
- **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**：工具权限边界不能在安装前相信。
- **继续会触碰**：命令执行、本地环境或项目文件、宿主 AI 上下文

### 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_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_0001` 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_0003` supported 0.86

### 现在还不能相信

- **工具权限边界不能在安装前相信。**（unverified）：MCP/tool 类项目通常会触碰文件、网络、浏览器或外部 API，必须真实检查权限和日志。
- **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。
- **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。
- **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`
- **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：`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。）
- **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。
- **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_0007` inferred 0.45
- **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_0008` 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

- **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_0001` supported 0.86

### 上下文规模

- 文件总数：157
- 重要文件覆盖：40/157
- 证据索引条目：60
- 角色 / Skill 条目：7

### 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 visionaire-engine 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 visionaire-engine 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 visionaire-engine 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 7 个角色 / Skill / 项目文档条目。

- **Visionaire Engine**（project_doc）：Which rule, which file, which line — and why it wins. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`README.md`
- **Architecture**（project_doc）：How visionaire-engine turns a rendered page into deterministic, attributable answers — and why it is built this way. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/architecture.md`
- **Install in your MCP client**（project_doc）：Visionaire is a standard stdio MCP server , so it works in any MCP-capable client — Claude, GitHub Copilot, Cursor, Google Antigravity, Windsurf, Zed, and others. The server is identical everywhere; only the registration differs. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/clients.md`
- **Development guide**（project_doc）：Building, testing, and extending visionaire-engine. For what the tools do, see tools.md tools.md ; for how the pipeline works internally and the authoritative design reference, see architecture.md architecture.md . 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/development.md`
- **Tool reference**（project_doc）：Visionaire Engine exposes 19 MCP tools: three session tools connect , navigate , set viewport and sixteen inspection tools — ten for the frozen moment what the page looks like and why , three for the time dimension get listeners , explain animations , record interaction — what happens and why , plus three added in v0.4: interact drive the UI into a state and leave it there , measure element sub-pixel glyph/text-ink… 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/tools.md`
- **WordPress origin resolution**（project_doc）：Visionaire's headline trick: on a WordPress site, it doesn't just tell you which CSS rule wins — it tells you where that rule actually lives and where to go edit it , even when the "file" Chrome shows you is a generated artifact nobody should ever open. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/wordpress.md`
- **Contributing to Visionaire Engine**（project_doc）：Bug reports, feature requests, and pull requests are welcome — open an issue first for anything larger than a small fix. Field reports from real debugging sessions are especially valued — a description of where a tool got clumsy or gave the wrong context has driven most releases so far. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`CONTRIBUTING.md`

## 证据索引

- 共索引 60 条证据。

- **Visionaire Engine**（documentation）：Which rule, which file, which line — and why it wins. 证据：`README.md`
- **Package**（package_manifest）：{ "name": "visionaire-engine", "version": "0.6.2", "description": "Deterministic rendered-page \u2192 LLM context MCP server: which rule, which file, which line \u2014 and why it wins.", "type": "module", "bin": { "visionaire-engine": "dist/index.js" }, "main": "dist/index.js", "engines": { "node": " =20" }, "scripts": { "build": "tsc -p tsconfig.json", "dev": "tsx src/index.ts", "test": "vitest run", "bench": "tsx bench/run.ts", "demo": "tsx scripts/demo.ts" }, "keywords": "mcp", "css", "devtools", "debugging", "wordpress", "llm" , "license": "Apache-2.0", "dependencies": { "@jridgewell/trace-mapping": "^0.3.31", "@modelcontextprotocol/sdk": "^1.29.0", "puppeteer-core": "^24.43.1", "zod":… 证据：`package.json`
- **Architecture**（documentation）：How visionaire-engine turns a rendered page into deterministic, attributable answers — and why it is built this way. 证据：`docs/architecture.md`
- **Install in your MCP client**（documentation）：Visionaire is a standard stdio MCP server , so it works in any MCP-capable client — Claude, GitHub Copilot, Cursor, Google Antigravity, Windsurf, Zed, and others. The server is identical everywhere; only the registration differs. 证据：`docs/clients.md`
- **Development guide**（documentation）：Building, testing, and extending visionaire-engine. For what the tools do, see tools.md tools.md ; for how the pipeline works internally and the authoritative design reference, see architecture.md architecture.md . 证据：`docs/development.md`
- **Tool reference**（documentation）：Visionaire Engine exposes 19 MCP tools: three session tools connect , navigate , set viewport and sixteen inspection tools — ten for the frozen moment what the page looks like and why , three for the time dimension get listeners , explain animations , record interaction — what happens and why , plus three added in v0.4: interact drive the UI into a state and leave it there , measure element sub-pixel glyph/text-ink centering , and evaluate the escape hatch for genuinely bespoke reads . All output is plain text plus one PNG for annotated screenshot , deterministic, and token-budgeted. Concepts and vocabulary come from architecture.md architecture.md ; this page documents the tools as impleme… 证据：`docs/tools.md`
- **WordPress origin resolution**（documentation）：Visionaire's headline trick: on a WordPress site, it doesn't just tell you which CSS rule wins — it tells you where that rule actually lives and where to go edit it , even when the "file" Chrome shows you is a generated artifact nobody should ever open. 证据：`docs/wordpress.md`
- **Contributing to Visionaire Engine**（documentation）：Bug reports, feature requests, and pull requests are welcome — open an issue first for anything larger than a small fix. Field reports from real debugging sessions are especially valued — a description of where a tool got clumsy or gave the wrong context has driven most releases so far. 证据：`CONTRIBUTING.md`
- **License**（source_file）：Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ 证据：`LICENSE`
- **Glama**（structured_config）：{ "$schema": "https://glama.ai/mcp/schemas/server.json", "maintainers": "mi60dev" } 证据：`glama.json`
- **Demo**（source_file）：import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { SessionManager } from '../src/session.js' import { explainStylesTool } from '../src/tools/explain-styles.js' import { pageSnapshotTool } from '../src/tools/page-snapshot.js' ⋮---- interface DemoArgs { url: string selector: string } ⋮---- function parseArgs argv: string : DemoArgs ⋮---- async function main : Promise 证据：`scripts/demo.ts`
- **Scripts**（source_file）：import { TraceMap, originalPositionFor } from '@jridgewell/trace-mapping' import { readFile } from 'node:fs/promises' import { fileURLToPath } from 'node:url' import type { CDPSession, Protocol } from 'puppeteer-core' import type { AuthoredPos, ResolvedScriptPos, ScriptInfo, ScriptRegistryLike, WpSheetMeta } from '../types.js' import { resolveWpOrigin, wpOriginToStyleOrigin } from './wordpress.js' ⋮---- export class ScriptRegistry implements ScriptRegistryLike ⋮---- async attach cdp: CDPSession : Promise ⋮---- private register ev: Protocol.Debugger.ScriptParsedEvent : void ⋮---- get scriptId: string : ScriptInfo undefined ⋮---- async resolvePosition scriptId: string, line: number, column: n… 证据：`src/attribution/scripts.ts`
- **Sourcemaps**（source_file）：import { TraceMap, originalPositionFor } from '@jridgewell/trace-mapping' import { readFile } from 'node:fs/promises' import { fileURLToPath } from 'node:url' import type { AuthoredPos, SheetInfo } from '../types.js' ⋮---- export async function resolveAuthoredPosition sheet: SheetInfo, line: number, column: number, : Promise ⋮---- export function clearSourceMapCache : void ⋮---- async function loadTraceMap sourceMapURL: string, sheetURL: string : Promise ⋮---- // Passing the map URL makes originalPositionFor return resolved source URLs. ⋮---- function decodeDataUrl url: string : string undefined ⋮---- function resolveMapUrl mapUrl: string, baseUrl: string : string undefined ⋮---- async func… 证据：`src/attribution/sourcemaps.ts`
- **Stylesheets**（source_file）：import type { CDPSession, Protocol } from 'puppeteer-core' import type { SheetInfo, StyleOrigin, StylesheetRegistryLike, WpSheetMeta } from '../types.js' import { pairAttributes } from '../uid.js' import { resolveWpOrigin, wpOriginToStyleOrigin } from './wordpress.js' ⋮---- function looksMinified sheet: SheetInfo : boolean ⋮---- / SPEC §5.2/§8.3: a minified sheet with no source map has untrustworthy line numbers — degrade 'line' → 'file' and say so " file … — minified, no map " . With a source map present, sourcemaps.ts restores authored positions instead. / function degradeMinified origin: StyleOrigin, sheet: SheetInfo : StyleOrigin ⋮---- export class StylesheetRegistry implements Styleshe… 证据：`src/attribution/stylesheets.ts`
- **Wordpress**（source_file）：import type { CDPSession } from 'puppeteer-core' import type { PlatformInfo, StyleOrigin, WpOrigin, WpOriginKind, WpSheetMeta } from '../types.js' ⋮---- type WpRule = meta: WpSheetMeta = WpOrigin undefined ⋮---- export function resolveWpOrigin meta: WpSheetMeta : WpOrigin undefined ⋮---- function wpContentRelative url: string : string undefined ⋮---- function urlPath url: string : string ⋮---- function themeLike label: string, meta: WpSheetMeta : Omit ⋮---- export function wpOriginToStyleOrigin wp: WpOrigin, meta: WpSheetMeta : StyleOrigin ⋮---- export function detectPlatform sheets: Array , docMarkers: { generator?: string; bodyClasses?: string }, : PlatformInfo ⋮---- interface ThemeStat {… 证据：`src/attribution/wordpress.ts`
- **Alignment**（source_file）：export interface AlignBox { uid: string identity: string x: number y: number w: number h: number } ⋮---- export interface AlignmentOptions { tolerance?: number gridUnit?: number dpr?: number } ⋮---- interface Cluster { mean: number members: number } ⋮---- function clusters values: number , tol: number : Cluster ⋮---- const r1 = v: number : string ⋮---- function edgeLine name: string, values: number , boxes: AlignBox , tol: number : string undefined ⋮---- function gapLines boxes: AlignBox , axis: 'x' 'y', tol: number : string ⋮---- function gridLines boxes: AlignBox , unit: number, tol: number : string ⋮---- function pixelSnapLines boxes: AlignBox , dpr: number : string ⋮---- export function… 证据：`src/engine/alignment.ts`
- **Animations**（source_file）：import type { AnimationCensusEntry, AnimationFinding } from '../types.js' ⋮---- export interface AnimationIssueInput { census: AnimationCensusEntry transitionProperty?: string transitionDurationsMs?: number computed: Map reducedMotionActive: boolean expectedProperty?: string } ⋮---- interface RuleCtx { computed: Map declared: Set active: Set } ⋮---- interface PropertyRule { rule: AnimationFinding 'rule' match: prop: string, ctx: RuleCtx = boolean reason: prop: string, ctx: RuleCtx = string fixHint: prop: string, ctx: RuleCtx = string undefined } ⋮---- export function findAnimationIssues input: AnimationIssueInput : AnimationFinding ⋮---- export function serializeAnimationsExpression : string 证据：`src/engine/animations.ts`
- **Box Model**（source_file）：import type { Protocol } from 'puppeteer-core' import type { Bounds, BoxSummary, ResolvedNode, ToolContext } from '../types.js' ⋮---- function round1 n: number : number ⋮---- function quadBounds quad: Protocol.DOM.Quad : Bounds ⋮---- function sideWidths inner: Protocol.DOM.Quad, outer: Protocol.DOM.Quad, : number, number, number, number ⋮---- export async function getBoxSummary ctx: ToolContext, node: ResolvedNode, : Promise 证据：`src/engine/box-model.ts`
- **Cascade**（source_file）：import type { Protocol } from 'puppeteer-core' import type { DeclarationInfo, DeclOriginType, LossReason, PropertyVerdict, Specificity, } from '../types.js' import { compareSpecificity, computeSpecificity } from './specificity.js' ⋮---- function stripImportant value: string : string ⋮---- function isInheritable property: string : boolean ⋮---- type BucketOrigin = 'user-agent' 'inline' 'author' ⋮---- function bucketOf origin: BucketOrigin, important: boolean : number ⋮---- interface Candidate { decl: DeclarationInfo distance: number order: number bucket: number } ⋮---- interface RuleCtx { selector?: string specificity?: Specificity layerChain?: string media?: string styleSheetId?: string } ⋮… 证据：`src/engine/cascade.ts`
- **Specificity**（source_file）：import type { Specificity } from '../types.js' ⋮---- export function computeSpecificity selector: string : Specificity ⋮---- export function compareSpecificity a: Specificity, b: Specificity : number ⋮---- function maxOfList list: string : Specificity ⋮---- function complexSpecificity sel: string : Specificity ⋮---- function isIdentStart ch: string : boolean ⋮---- function skipIdent s: string, i: number : number ⋮---- / From an opening bracket, return the index just past its balanced close. Quote- and escape-aware. / function skipBalanced s: string, openIdx: number, open: string, close: string : number ⋮---- function skipString s: string, quoteIdx: number : number ⋮---- / Split on a separat… 证据：`src/engine/specificity.ts`
- **Check Alignment**（source_file）：import { z } from 'zod' import type { ToolContext, ToolDef } from '../types.js' import { sanitizePageText } from '../types.js' import { analyzeAlignment, type AlignBox } from '../engine/alignment.js' ⋮---- interface GatheredBox { identity: string x: number y: number w: number h: number } ⋮---- async handler ctx, args 证据：`src/tools/check-alignment.ts`
- **Explain Animations**（source_file）：import { z } from 'zod' import type { Protocol } from 'puppeteer-core' import type { AnimationCensusEntry, AuthoredPos, ElementSummary, StyleOrigin, ToolContext, ToolDef, ToolResult, } from '../types.js' import { estimateTokens } from '../types.js' import { pairAttributes, resolveTarget } from '../uid.js' import { computeCascade } from '../engine/cascade.js' import { findAnimationIssues, serializeAnimationsExpression } from '../engine/animations.js' import { resolveAuthoredPosition } from '../attribution/sourcemaps.js' ⋮---- interface ExplainAnimationsArgs { uid?: string selector?: string x?: number y?: number property?: string } ⋮---- async function explainAnimations ctx: ToolContext, args… 证据：`src/tools/explain-animations.ts`
- **Explain Styles**（source_file）：import { z } from 'zod' import type { Protocol } from 'puppeteer-core' import type { AttributedVerdict, DeclarationInfo, ElementSummary, PropertyVerdict, Specificity, ToolContext, ToolDef, ToolResult, VisibilityReport, WhyDossierInput, } from '../types.js' import { estimateTokens } from '../types.js' import { buildScopeNotes, collectScopeData } from '../engine/scope.js' import { pairAttributes, resolveTarget } from '../uid.js' import { computeCascade } from '../engine/cascade.js' import { findInactiveDeclarations } from '../engine/inactive.js' import { assessVisibility } from '../engine/visibility.js' import { resolveAuthoredPosition } from '../attribution/sourcemaps.js' import { renderWhyD… 证据：`src/tools/explain-styles.ts`
- **Find Elements**（source_file）：import type { Protocol } from 'puppeteer-core' import { z } from 'zod' import type { ToolContext, ToolDef } from '../types.js' import { sanitizePageText } from '../types.js' import { selectorHelp } from '../engine/suggest.js' ⋮---- interface MatchItem { tag: string classes: string id?: string text?: string href?: string x: number y: number w: number h: number } ⋮---- function searchExpression crit: z.infer : string ⋮---- function describeException details: Protocol.Runtime.ExceptionDetails : string ⋮---- function identityOf item: ⋮---- async function unpackNodeArray ctx: ToolContext, arrayObjectId: string, : Promise : number ⋮---- // A guessed selector that matches nothing is the common fai… 证据：`src/tools/find-elements.ts`
- **Page Snapshot**（source_file）：import type { Protocol } from 'puppeteer-core' import { z } from 'zod' import type { Bounds, PageMeta, SnapshotNode, TargetSpec, ToolContext, ToolDef, ToolResult, } from '../types.js' import { COMPUTED WHITELIST } from '../types.js' import { resolveTarget } from '../uid.js' import { renderCensus } from '../format/census.js' import { detectPlatformFromPage } from '../attribution/wordpress.js' ⋮---- function snapshotToTree ctx: ToolContext, doc: Protocol.DOMSnapshot.DocumentSnapshot, strings: string , viewport: Bounds, scopeBackendId?: number, : SnapshotNode ⋮---- const str = i: number undefined : string undefined ⋮---- const tagOf = i: number : string ⋮---- // Pseudo-element rows ::before et… 证据：`src/tools/page-snapshot.ts`
- **Pick Color**（source_file）：import { z } from 'zod' import type { ToolContext, ToolDef } from '../types.js' import { pairAttributes, resolveTarget } from '../uid.js' import { decodePng } from '../engine/png.js' import { contrastRatio, contrastVerdict, parseCssColor, toHex, type Rgba } from '../engine/color.js' ⋮---- async handler ctx, args 证据：`src/tools/pick-color.ts`
- **Manifest**（structured_config）：{ "id": 1, "file": "cases/case01-specificity-override.html", "user report": "I set the subscribe button's bottom margin to 8px but it still has a huge gap under it.", "target": { "selector": ".btn" }, "tool": "explain styles", "args": { "property": "margin-bottom" }, "expected": "WINNER", "32px", "case01.css:10", "lost specificity ", "8px", "case01.css:14" }, { "id": 2, "file": "cases/case02-source-order.html", "user report": "The nav links should be blue per my stylesheet but they render grey.", "target": { "selector": ".nav a" }, "tool": "explain styles", "args": { "property": "color" }, "expected": "WINNER", " 777777", "case02-skin.css:10", "lost order ", " 0055ff", "case02-base.css:10"… 证据：`bench/manifest.json`
- **Tsconfig**（structured_config）：{ "compilerOptions": { "target": "ES2022", "module": "NodeNext", "moduleResolution": "NodeNext", "lib": "ES2022", "DOM" , "strict": true, "esModuleInterop": true, "skipLibCheck": true, "declaration": true, "sourceMap": true, "outDir": "dist", "rootDir": "src", "resolveJsonModule": true, "forceConsistentCasingInFileNames": true }, "include": "src" } 证据：`tsconfig.json`
- **.dockerignore**（source_file）：node modules dist test bench launch scripts .github .git .tgz 证据：`.dockerignore`
- **internal design/development doc — not published**（source_file）：internal design/development doc — not published SPEC.md launch/ 证据：`.gitignore`
- **syntax=docker/dockerfile:1**（source_file）：syntax=docker/dockerfile:1 Build stage: compile TypeScript with dev deps, then prune to production. FROM node:20-slim AS build WORKDIR /app COPY package.json package-lock.json ./ RUN npm ci COPY tsconfig.json ./ COPY src ./src RUN npm run build && npm prune --omit=dev 证据：`Dockerfile`
- **Notice**（source_file）：Visionaire Engine Copyright 2026 Mihail Valchev https://github.com/mi60dev/visionaire-engine 证据：`NOTICE`
- **Run**（source_file）：import fs from 'node:fs' import http from 'node:http' import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { findChromeExecutable, SessionManager } from '../src/session.js' import { explainStylesTool } from '../src/tools/explain-styles.js' import { inspectAncestorsTool } from '../src/tools/inspect-ancestors.js' import { inspectElementTool } from '../src/tools/inspect-element.js' import { pageSnapshotTool } from '../src/tools/page-snapshot.js' import type { ToolDef } from '../src/types.js' import { estimateTokens } from '../src/types.js' ⋮---- async function resolveTool name: string : Promise ⋮---- interface BenchCase { id: number file: string user repo… 证据：`bench/run.ts`
- **Index**（source_file）：import { StdioServerTransport } from '@modelcontextprotocol/sdk/server/stdio.js' import { createServer } from './server.js' import { SessionManager } from './session.js' ⋮---- async function shutdown signal: string : Promise 证据：`src/index.ts`
- **Server**（source_file）：import { createRequire } from 'node:module' import { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js' ⋮---- import type { CallToolResult } from '@modelcontextprotocol/sdk/types.js' import { z } from 'zod' import type { SessionManager } from './session.js' import type { ToolDef, ToolResult } from './types.js' import { annotatedScreenshotTool } from './tools/annotated-screenshot.js' import { checkAlignmentTool } from './tools/check-alignment.js' import { evaluateTool } from './tools/evaluate.js' import { pickColorTool } from './tools/pick-color.js' import { explainAnimationsTool } from './tools/explain-animations.js' import { explainStylesTool } from './tools/explain-styles.js' impo… 证据：`src/server.ts`
- **Session**（source_file）：import { spawnSync } from 'node:child process' import fs from 'node:fs' import os from 'node:os' import path from 'node:path' import puppeteer from 'puppeteer-core' import type { Browser, CDPSession, LaunchOptions, Page, Protocol } from 'puppeteer-core' import { ScriptRegistry } from './attribution/scripts.js' import { StylesheetRegistry } from './attribution/stylesheets.js' import type { ToolContext } from './types.js' import { UidRegistry } from './uid.js' ⋮---- export function findPuppeteerCachedChrome baseDir: string : string undefined ⋮---- function puppeteerCacheChromeDirs : string ⋮---- async function enableDebugger cdp: CDPSession : Promise ⋮---- export function findChromeExecutable… 证据：`src/session.ts`
- **Uid**（source_file）：import type { Protocol } from 'puppeteer-core' import type { ResolvedNode, TargetSpec, ToolContext, UidEntry, UidRegistryLike } from './types.js' import { sanitizePageText } from './types.js' import { selectorHelp } from './engine/suggest.js' ⋮---- function sanitizeMeta meta: Partial , : Partial ⋮---- export class UidRegistry implements UidRegistryLike ⋮---- assign backendNodeId: number, meta: Partial = ⋮---- get uid: string : UidEntry undefined ⋮---- byBackendId backendNodeId: number : string undefined ⋮---- clear : void ⋮---- export async function resolveTarget ctx: ToolContext, target: TargetSpec : Promise ⋮---- export function pairAttributes flat: string undefined : Map 证据：`src/uid.ts`
- **Animations.E2E.Test**（source_file）：import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { afterAll, beforeAll, describe, expect, it } from 'vitest' import { findChromeExecutable, SessionManager } from '../src/session.js' import { explainAnimationsTool } from '../src/tools/explain-animations.js' ⋮---- const fixtureUrl = name: string : string ⋮---- async function explain selector: string : Promise 证据：`test/animations.e2e.test.ts`
- **Animations.Test**（source_file）：import { describe, expect, it } from 'vitest' import { COMPOSITOR FRIENDLY PROPS, findAnimationIssues, NON ANIMATABLE PROPS, serializeAnimationsExpression, type AnimationIssueInput, } from '../src/engine/animations.js' import type { AnimationCensusEntry, AnimationFinding } from '../src/types.js' ⋮---- function entry partial: Partial = ⋮---- function run partial: Partial = ⋮---- const ofRule = findings: AnimationFinding , rule: AnimationFinding 'rule' : AnimationFinding 证据：`test/animations.test.ts`
- **Blast.E2E.Test**（source_file）：import { describe, it, expect, beforeAll, afterAll } from 'vitest' import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { SessionManager, findChromeExecutable } from '../src/session.js' import { explainStylesTool } from '../src/tools/explain-styles.js' import type { ToolContext } from '../src/types.js' 证据：`test/blast.e2e.test.ts`
- **Cascade.Test**（source_file）：import { describe, expect, it } from 'vitest' import type { Protocol } from 'puppeteer-core' import { computeCascade } from '../src/engine/cascade.js' import { compareSpecificity, computeSpecificity } from '../src/engine/specificity.js' import type { PropertyVerdict } from '../src/types.js' ⋮---- type CSSProperty = Protocol.CSS.CSSProperty type CSSStyle = Protocol.CSS.CSSStyle type RuleMatch = Protocol.CSS.RuleMatch type Matched = Protocol.CSS.GetMatchedStylesForNodeResponse ⋮---- function prop name: string, value: string, extra: Partial = ⋮---- function style props: CSSProperty , extra: Partial = ⋮---- function ruleMatch selector: string, props: CSSProperty , extra: { origin?: Protocol.CSS… 证据：`test/cascade.test.ts`
- **Coldstart.Test**（source_file）：import { afterEach, beforeEach, describe, expect, it } from 'vitest' import fs from 'node:fs' import os from 'node:os' import path from 'node:path' import { findPuppeteerCachedChrome, isSandboxBlocked } from '../src/session.js' ⋮---- function relBinary : string ⋮---- function platformDir : string ⋮---- function writeFakeChrome base: string, version: string : string 证据：`test/coldstart.test.ts`
- **E2E.Test**（source_file）：import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { afterAll, beforeAll, describe, expect, it } from 'vitest' import { findChromeExecutable, SessionManager } from '../src/session.js' import { explainStylesTool } from '../src/tools/explain-styles.js' import { inspectAncestorsTool } from '../src/tools/inspect-ancestors.js' import { inspectElementTool } from '../src/tools/inspect-element.js' import { pageOriginsTool } from '../src/tools/page-origins.js' import { pageSnapshotTool } from '../src/tools/page-snapshot.js' import { resolveTarget } from '../src/uid.js' ⋮---- const fixtureUrl = name: string : string ⋮---- function uidFrom dossierText: string :… 证据：`test/e2e.test.ts`
- **Evaluate.E2E.Test**（source_file）：import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { afterAll, beforeAll, describe, expect, it } from 'vitest' import { findChromeExecutable, SessionManager } from '../src/session.js' import { evaluateTool } from '../src/tools/evaluate.js' import type { ToolContext } from '../src/types.js' ⋮---- async function run args: Record : Promise 证据：`test/evaluate.e2e.test.ts`
- **Find.E2E.Test**（source_file）：import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { afterAll, beforeAll, describe, expect, it } from 'vitest' import { findChromeExecutable, SessionManager } from '../src/session.js' import { findElementsTool } from '../src/tools/find-elements.js' import type { ToolContext } from '../src/types.js' ⋮---- const run = args: Record : Promise 证据：`test/find.e2e.test.ts`
- **Geometry.Test**（source_file）：import { describe, expect, it } from 'vitest' import { centeringDeltas, inkBox, measureExpression, type ContentBox, type InkMetrics, } from '../src/engine/geometry.js' ⋮---- function box cx: number, cy: number, w: number, h: number : ContentBox ⋮---- function inkCenteredAt cx: number, cy: number, hw: number, hh: number : InkMetrics 证据：`test/geometry.test.ts`
- **Hardening.E2E.Test**（source_file）：import { describe, it, expect, beforeAll, afterAll } from 'vitest' import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { SessionManager, findChromeExecutable } from '../src/session.js' import { pageSnapshotTool } from '../src/tools/page-snapshot.js' import { inspectElementTool } from '../src/tools/inspect-element.js' import type { ToolContext } from '../src/types.js' 证据：`test/hardening.e2e.test.ts`
- **Inactive.Test**（source_file）：import { describe, expect, it } from 'vitest' import { findInactiveDeclarations } from '../src/engine/inactive.js' import type { DeclarationInfo } from '../src/types.js' ⋮---- function decl property: string, value: string : DeclarationInfo ⋮---- function cm entries: Record = 证据：`test/inactive.test.ts`
- **Inject.E2E.Test**（source_file）：import { describe, it, expect, beforeAll, afterAll } from 'vitest' import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { SessionManager, findChromeExecutable } from '../src/session.js' import { injectCssTool } from '../src/tools/inject-css.js' import { evaluateTool } from '../src/tools/evaluate.js' import type { ToolContext } from '../src/types.js' ⋮---- async function computedOf ctx: ToolContext, selector: string, prop: string : Promise 证据：`test/inject.e2e.test.ts`
- **Interact.E2E.Test**（source_file）：import fs from 'node:fs' import http from 'node:http' import path from 'node:path' import { fileURLToPath } from 'node:url' import type { AddressInfo } from 'node:net' import { afterAll, beforeAll, describe, expect, it } from 'vitest' import { findChromeExecutable, SessionManager } from '../src/session.js' import { interactTool } from '../src/tools/interact.js' import { inspectElementTool } from '../src/tools/inspect-element.js' 证据：`test/interact.e2e.test.ts`
- **Interaction.E2E.Test**（source_file）：import fs from 'node:fs' import http from 'node:http' import path from 'node:path' import { fileURLToPath } from 'node:url' import type { AddressInfo } from 'node:net' import { afterAll, beforeAll, describe, expect, it } from 'vitest' import { buildTimeline, renderTimeline } from '../src/engine/timeline.js' import { findChromeExecutable, SessionManager } from '../src/session.js' import { recordInteractionTool } from '../src/tools/record-interaction.js' import type { TimelineEvent } from '../src/types.js' ⋮---- const sleep = ms: number : Promise ⋮---- const ev = partial: Partial & ⋮---- const listener = : void = 证据：`test/interaction.e2e.test.ts`
- **Listeners.E2E.Test**（source_file）：import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { afterAll, beforeAll, describe, expect, it } from 'vitest' import { findChromeExecutable, SessionManager } from '../src/session.js' import { getListenersTool } from '../src/tools/get-listeners.js' ⋮---- const fixtureUrl = name: string : string ⋮---- async function run args: Record : Promise 证据：`test/listeners.e2e.test.ts`
- **Measure.E2E.Test**（source_file）：import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { afterAll, beforeAll, describe, expect, it } from 'vitest' import { findChromeExecutable, SessionManager } from '../src/session.js' import { measureElementTool } from '../src/tools/measure-element.js' ⋮---- const fixtureUrl = name: string : string ⋮---- function delta text: string, axis: 'horizontal' 'vertical' : number ⋮---- function dims text: string, label: string : ⋮---- async function run args: Record : Promise 证据：`test/measure.e2e.test.ts`
- **Pick.E2E.Test**（source_file）：import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { afterAll, beforeAll, describe, expect, it } from 'vitest' import { findChromeExecutable, SessionManager } from '../src/session.js' import { pickElementTool } from '../src/tools/pick-element.js' ⋮---- const fixtureUrl = name: string : string ⋮---- const sleep = ms: number : Promise ⋮---- const listener = : void = 证据：`test/pick.e2e.test.ts`
- **Pixel.E2E.Test**（source_file）：import { describe, it, expect, beforeAll, afterAll } from 'vitest' import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { SessionManager, findChromeExecutable } from '../src/session.js' import { checkAlignmentTool } from '../src/tools/check-alignment.js' import { pickColorTool } from '../src/tools/pick-color.js' import type { ToolContext } from '../src/types.js' 证据：`test/pixel.e2e.test.ts`
- **Pixel.Test**（source_file）：import { describe, it, expect } from 'vitest' import zlib from 'node:zlib' import { analyzeAlignment, type AlignBox } from '../src/engine/alignment.js' import { contrastRatio, contrastVerdict, parseCssColor, relativeLuminance, toHex } from '../src/engine/color.js' import { decodePng } from '../src/engine/png.js' ⋮---- const box = uid: string, x: number, y: number, w = 80, h = 40 : AlignBox = ⋮---- function tinyPng : Buffer ⋮---- const chunk = type: string, data: Buffer : Buffer = 证据：`test/pixel.test.ts`
- **Screenshot.E2E.Test**（source_file）：import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { afterAll, beforeAll, describe, expect, it } from 'vitest' import { findChromeExecutable, SessionManager } from '../src/session.js' import { annotatedScreenshotTool } from '../src/tools/annotated-screenshot.js' import { resolveTarget } from '../src/uid.js' import type { ToolResult } from '../src/types.js' ⋮---- const fixtureUrl = name: string : string ⋮---- function pngSize base64: string : ⋮---- function onlyImage res: ToolResult : ⋮---- const run = args: Record : Promise 证据：`test/screenshot.e2e.test.ts`
- **Suggest.E2E.Test**（source_file）：import { describe, it, expect, beforeAll, afterAll } from 'vitest' import path from 'node:path' import { fileURLToPath, pathToFileURL } from 'node:url' import { SessionManager, findChromeExecutable } from '../src/session.js' import { inspectElementTool } from '../src/tools/inspect-element.js' import { findElementsTool } from '../src/tools/find-elements.js' import type { ToolContext } from '../src/types.js' 证据：`test/suggest.e2e.test.ts`
- **Wordpress.Test**（source_file）：import type { Protocol } from 'puppeteer-core' import { describe, expect, it } from 'vitest' import { detectPlatform, resolveWpOrigin, wpOriginToStyleOrigin, } from '../src/attribution/wordpress.js' import { StylesheetRegistry } from '../src/attribution/stylesheets.js' import type { SheetInfo, WpSheetMeta } from '../src/types.js' ⋮---- function meta partial: Partial & ⋮---- // URL mentions both /cache/min/ and a theme path — optimizer must win by table order. ⋮---- { sourceURL: '' }, // inline sheet: empty sourceURL must not break detection ⋮---- function sheetInfo opts: { sourceURL: string length?: number endLine?: number sourceMapURL?: string } : SheetInfo 证据：`test/wordpress.test.ts`
- **Vitest.Config**（source_file）：import { defineConfig } from 'vitest/config' 证据：`vitest.config.ts`

## 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：`README.md`, `package.json`, `docs/architecture.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：`README.md`, `package.json`, `docs/architecture.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

下面内容用于强化 Repomix/AI Context Pack 主体。Human Manual 只提供阅读骨架；踩坑日志会被转成宿主 AI 必须遵守的工作约束。

## Human Manual 骨架

使用规则：这里只是项目阅读路线和显著性信号，不是事实权威。具体事实仍必须回到 repo evidence / Claim Graph。

宿主 AI 硬性规则：
- 不得把页标题、章节顺序、摘要或 importance 当作项目事实证据。
- 解释 Human Manual 骨架时，必须明确说它只是阅读路线/显著性信号。
- 能力、安装、兼容性、运行状态和风险判断必须引用 repo evidence、source path 或 Claim Graph。

- **项目概览与快速开始**：importance `high`
  - source_paths: README.md, package.json, docs/clients.md, glama.json, Dockerfile
- **22 个 MCP 工具完整参考**：importance `high`
  - source_paths: docs/tools.md, src/tools/page-snapshot.ts, src/tools/explain-styles.ts, src/tools/find-elements.ts, src/tools/check-alignment.ts
- **确定性 CSS 分析引擎架构**：importance `high`
  - source_paths: docs/architecture.md, src/engine/cascade.ts, src/engine/specificity.ts, src/engine/box-model.ts, src/engine/animations.ts
- **WordPress 归属解析与生产部署**：importance `high`
  - source_paths: docs/wordpress.md, docs/development.md, src/attribution/wordpress.ts, src/attribution/stylesheets.ts, src/attribution/scripts.ts

## Repo Inspection Evidence / 源码检查证据

- repo_clone_verified: true
- repo_inspection_verified: true
- repo_commit: `326e13340956d5f6ee40309ba831e00dc3819577`
- inspected_files: `Dockerfile`, `README.md`, `package.json`, `docs/architecture.md`, `docs/clients.md`, `docs/development.md`, `docs/tools.md`, `docs/wordpress.md`, `src/attribution/scripts.ts`, `src/attribution/sourcemaps.ts`, `src/attribution/stylesheets.ts`, `src/attribution/wordpress.ts`, `src/engine/alignment.ts`, `src/engine/ancestors.ts`, `src/engine/animations.ts`, `src/engine/box-model.ts`, `src/engine/cascade.ts`, `src/engine/color.ts`, `src/engine/geometry.ts`, `src/engine/inactive.ts`

宿主 AI 硬性规则：
- 没有 repo_clone_verified=true 时，不得声称已经读过源码。
- 没有 repo_inspection_verified=true 时，不得把 README/docs/package 文件判断写成事实。
- 没有 quick_start_verified=true 时，不得声称 Quick Start 已跑通。

## Doramagic Pitfall Constraints / 踩坑约束

这些规则来自 Doramagic 发现、验证或编译过程中的项目专属坑点。宿主 AI 必须把它们当作工作约束，而不是普通说明文字。

### Constraint 1: 可能修改宿主 AI 配置

- Trigger: 项目面向 Claude/Cursor/Codex/Gemini/OpenCode 等宿主，或安装命令涉及用户配置目录。
- Host AI rule: 列出会写入的配置文件、目录和卸载/回滚步骤。
- Why it matters: 安装可能改变本机 AI 工具行为，用户需要知道写入位置和回滚方法。
- Evidence: capability.host_targets | https://github.com/mi60dev/visionaire-engine | host_targets=mcp_host, claude
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 2: 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 | https://github.com/mi60dev/visionaire-engine | README/documentation is current enough for a first validation pass.
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 3: 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 | https://github.com/mi60dev/visionaire-engine | last_activity_observed missing
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

- Trigger: no_demo
- Evidence: downstream_validation.risk_items | https://github.com/mi60dev/visionaire-engine | no_demo; severity=medium
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 5: 存在评分风险

- Trigger: no_demo
- Why it matters: 风险会影响是否适合普通用户安装。
- Evidence: risks.scoring_risks | https://github.com/mi60dev/visionaire-engine | no_demo; severity=medium
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 6: issue/PR 响应质量未知

- Trigger: issue_or_pr_quality=unknown。
- Host AI rule: 抽样最近 issue/PR，判断是否长期无人处理。
- Why it matters: 用户无法判断遇到问题后是否有人维护。
- Evidence: evidence.maintainer_signals | https://github.com/mi60dev/visionaire-engine | issue_or_pr_quality=unknown
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 7: 发布节奏不明确

- Trigger: release_recency=unknown。
- Host AI rule: 确认最近 release/tag 和 README 安装命令是否一致。
- Why it matters: 安装命令和文档可能落后于代码，用户踩坑概率升高。
- Evidence: evidence.maintainer_signals | https://github.com/mi60dev/visionaire-engine | release_recency=unknown
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。
