# m1nd - 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 m1nd 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

- **AI 研究者或研究型 Agent 构建者**：README 明确围绕研究、实验或论文工作流展开。 证据：`README.md` Claim：`clm_0003` supported 0.86
- **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：`skills/m1nd-first/SKILL.md`, `skills/m1nd-guardian/SKILL.md`, `skills/m1nd-operator/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：`skills/m1nd-first/SKILL.md`, `skills/m1nd-guardian/SKILL.md`, `skills/m1nd-operator/SKILL.md` Claim：`clm_0001` 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`, `docs/wiki/src/tutorials/quickstart.md` Claim：`clm_0002` supported 0.86

## 怎么开始

- `npx -y @maxkle1nz/m1nd update apply --yes` 证据：`README.md` Claim：`clm_0006` supported 0.86
- `npx -y @maxkle1nz/m1nd doctor` 证据：`README.md` Claim：`clm_0007` supported 0.86
- `npx -y @maxkle1nz/m1nd hosts apply --host claude --project . --yes` 证据：`README.md` Claim：`clm_0008` supported 0.86
- `npx -y @maxkle1nz/m1nd agent first-minute --repo . --query "map this repo" --json` 证据：`README.md` Claim：`clm_0009` supported 0.86
- `git clone https://github.com/maxkle1nz/m1nd.git` 证据：`docs/wiki/src/tutorials/quickstart.md` Claim：`clm_0010` unverified 0.25

## 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、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：AI 研究者或研究型 Agent 构建者**（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_0003` supported 0.86
- **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：`skills/m1nd-first/SKILL.md`, `skills/m1nd-guardian/SKILL.md`, `skills/m1nd-operator/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：`skills/m1nd-first/SKILL.md`, `skills/m1nd-guardian/SKILL.md`, `skills/m1nd-operator/SKILL.md` Claim：`clm_0001` 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`, `docs/wiki/src/tutorials/quickstart.md` Claim：`clm_0002` 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

- **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。 证据：`AGENTS.md`, `skills/m1nd-first/SKILL.md`, `skills/m1nd-guardian/SKILL.md`, `skills/m1nd-operator/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。
- **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`, `docs/wiki/src/tutorials/quickstart.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：`AGENTS.md`, `skills/m1nd-first/SKILL.md`, `skills/m1nd-guardian/SKILL.md`, `skills/m1nd-operator/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：`README.md`, `docs/wiki/src/tutorials/quickstart.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_0011` 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`, `docs/wiki/src/tutorials/quickstart.md` Claim：`clm_0012` 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：`skills/m1nd-first/SKILL.md`, `skills/m1nd-guardian/SKILL.md`, `skills/m1nd-operator/SKILL.md` Claim：`clm_0001` 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`, `docs/wiki/src/tutorials/quickstart.md` Claim：`clm_0002` supported 0.86

### 上下文规模

- 文件总数：895
- 重要文件覆盖：40/895
- 证据索引条目：80
- 角色 / Skill 条目：3

### 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 m1nd 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 m1nd 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 m1nd 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 3 个角色 / Skill / 项目文档条目。

- **m1nd-first**（skill）：Use when investigating a repository, searching for implementation, reviewing changes, working from specs/docs, or preparing a risky code change in an environment where m1nd is available. This doctrine makes m1nd the first investigative layer before grep, glob, or manual file reads, except when the task is pure compiler/runtime truth or the exact file and lines are already known. 激活提示：当用户任务与“m1nd-first”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：`skills/m1nd-first/SKILL.md`
- **m1nd-guardian**（skill）：Use when continuing, coordinating, reviewing, or handing off the M1ND-10 program; protecting its candidate tree; advancing G0-G10; freezing or releasing a candidate; activating autonomy; or when the user asks an agent to act as M1ND guardian. Enforces one-active-front execution, proof-state separation, same-candidate receipts, budget and stop rules, succession testing, and human-only sovereignty. Do not use for ordi… 激活提示：当用户任务与“m1nd-guardian”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：`skills/m1nd-guardian/SKILL.md`
- **m1nd-operator**（skill）：Use when the user mentions m1nd or when repo investigation, search, review, docs/spec work, or risky change prep should go through m1nd first before grep, glob, or manual file reads. Covers m1nd-first routing, L1GHT and universal document ingestion, risky edit preparation, document-to-code binding, the served owner + --attach bridge transport, the delegation layer delegate/debrief packets , medulla tiers and audited… 激活提示：当用户任务与“m1nd-operator”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：`skills/m1nd-operator/SKILL.md`

## 证据索引

- 共索引 80 条证据。

- **Docs guide**（documentation）：This repo keeps the public docs surface intentionally small. 证据：`docs/README.md`
- **Benchmark Harness**（documentation）：This directory holds reproducible benchmark inputs for m1nd . 证据：`docs/benchmarks/README.md`
- **The audit your agent stops paying for**（documentation）：m1nd gives your coding agent a brain per repository: a local code graph served over MCP, memory anchored to the code it cites, and a trust verdict on every answer. "Insufficient evidence" is a real answer here. So is "don't trust this yet, and here is how to repair it". 证据：`README.md`
- **m1nd Agent Skills**（documentation）：This directory is the installable agent doctrine for m1nd . 证据：`skills/README.md`
- **Package**（package_manifest）：{ "name": "m1nd-demo", "private": true, "version": "0.2.0", "type": "module", "scripts": { "dev": "vite --config vite.config.ts --host 0.0.0.0 --port 5175", "build": "tsc --noEmit && vite build --config vite.config.ts && cp dist/index.html dist/404.html && mkdir -p dist/use-cases dist/demo dist/l1ght && cp dist/index.html dist/use-cases/index.html && cp dist/index.html dist/demo/index.html && cp dist/index.html dist/l1ght/index.html", "preview": "vite preview --config vite.config.ts" }, "dependencies": { "@radix-ui/react-accordion": "^1.2.4", "@radix-ui/react-alert-dialog": "^1.1.7", "@radix-ui/react-aspect-ratio": "^1.1.3", "@radix-ui/react-avatar": "^1.1.4", "@radix-ui/react-checkbox": "^… 证据：`m1nd-demo/package.json`
- **Package**（package_manifest）：{ "name": "m1nd-ui", "private": true, "version": "0.1.0", "type": "module", "scripts": { "dev": "vite", "build": "tsc -b && vite build", "preview": "vite preview", "lint": "eslint .", "violet-lint": "node scripts/violet-lint.mjs", "icon-lint": "node scripts/icon-lint.mjs", "lint:soft": "node scripts/violet-lint.mjs && node scripts/icon-lint.mjs", "test": "node --import tsx --test --test-reporter=spec 'src/ / .test.ts' 'src/ / .test.tsx'", "test:e2e": "playwright test", "test:e2e:a11y": "playwright test --config playwright.a11y.config.ts", "test:e2e:live:contract": "node --import tsx --test --test-reporter=spec 'e2e-live/ .test.ts'", "test:e2e:live": "npm run test:e2e:live:contract && playwr… 证据：`m1nd-ui/package.json`
- **Package**（package_manifest）：{ "name": "@maxkle1nz/m1nd", "version": "1.5.0", "mcpName": "io.github.maxkle1nz/m1nd", "description": "Universal installer and agent pack for the m1nd MCP runtime.", "license": "MIT", "repository": { "type": "git", "url": "git+https://github.com/maxkle1nz/m1nd.git" }, "homepage": "https://github.com/maxkle1nz/m1nd readme", "bugs": { "url": "https://github.com/maxkle1nz/m1nd/issues" }, "type": "commonjs", "bin": { "m1nd": "npm/bin/m1nd.js", "m1nd-north-shim": "npm/bin/m1nd-north-shim.js" }, "files": "npm/bin/", "npm/lib/", "skills/README.md", "skills/m1nd-universal-agent-pack.md", "skills/m1nd-first/SKILL.md", "skills/m1nd-guardian/SKILL.md", "skills/m1nd-operator/SKILL.md", "skills/m1nd-op… 证据：`package.json`
- **AGENTS.md — working guide for any coding agent on m1nd**（documentation）：AGENTS.md — working guide for any coding agent on m1nd 证据：`AGENTS.md`
- **Contributing to m1nd**（documentation）：Thanks for your interest in contributing to m1nd. This document covers the basics. 证据：`CONTRIBUTING.md`
- **m1nd-first**（skill_instruction）：- Start with m1nd . In an MCP session, the front door is one call: north task — it composes trust, task context, prior cross-session memory, a sufficiency signal, one next move , and honest gaps into a single packet before you read or edit anything. If it returns needs ingest empty/unbound graph , do not call generic ingest : that mutation surface is policy-disabled. Use the isolated m1nd agent ... --repo CLI path for investigation, or the exact authority flow plus external mutation service for a governed existing-brain mutation. - Before rg , shell globbing, or manual file reads, m1nd answers or narrows FIRST — that is the default, not a question you ask only when convenient. The one excep… 证据：`skills/m1nd-first/SKILL.md`
- **m1nd-guardian**（skill_instruction）：This skill turns M1ND-10 continuation into a proof-carrying, resumable process. It complements m1nd-operator ; it does not replace repository law, frozen contracts, the current handoff, direct proof, or owner authority. 证据：`skills/m1nd-guardian/SKILL.md`
- **m1nd Operator**（skill_instruction）：This is the deep execution manual that complements the short m1nd-first doctrine. 证据：`skills/m1nd-operator/SKILL.md`
- **License**（source_file）：Copyright c 2026 Max Elias Kleinschmidt -- COSMOPHONIX INTELLIGENCE 证据：`LICENSE`
- **Agent-First Demo**（documentation）：This page is the shortest honest way to see why m1nd matters for coding agents. 证据：`docs/AGENT-FIRST-DEMO.md`
- **Agent Packs**（documentation）：m1nd ships an installable agent doctrine, not only an MCP binary. 证据：`docs/AGENT-PACKS.md`
- **Agent Tasknotes**（documentation）：Operational notes from real agent usage of m1nd . 证据：`docs/AGENT-TASKNOTES.md`
- **Case Intelligence — PRD**（documentation）：The mailbox learns to think · fingerprints, cases, absence sentinels, claim-vs-measure, and the abandonment signal — five organs, one law: a case is an overlay, never an axis 证据：`docs/CASE-INTELLIGENCE-PRD.md`
- **m1nd Focus Runtime — Goal, Process UML & PRD**（documentation）：m1nd Focus Runtime — Goal, Process UML & PRD 证据：`docs/FOCUS-RUNTIME-PRD.md`
- **GENESIS — Ingest Consumers Spec v2, post-verdict**（documentation）：GENESIS — Ingest Consumers Spec v2, post-verdict 证据：`docs/GENESIS-INGEST-CONSUMERS-SPEC.md`
- **HOST-INTEGRATION-MATRIX — the ambient layer on every agent host**（documentation）：HOST-INTEGRATION-MATRIX — the ambient layer on every agent host 证据：`docs/HOST-INTEGRATION-MATRIX.md`
- **m1nd Human Layer — PRD**（documentation）：The Living Tree · the Post-it Memory · the Pre-Flight Hero vision → spec 证据：`docs/HUMAN-LAYER-PRD.md`
- **Human View v2 — F0 technical amendment**（documentation）：Human View v2 — F0 technical amendment 证据：`docs/HUMAN-VIEW-V2-F0-TECH.md`
- **HUMAN VIEW v2 — F0c technical amendment: skeleton candidate — any repo gets a map**（documentation）：HUMAN VIEW v2 — F0c technical amendment: skeleton candidate — any repo gets a map 证据：`docs/HUMAN-VIEW-V2-F0C-TECH.md`
- **HUMAN VIEW v2 — F11 technical amendment: candidate editing with minimum human friction**（documentation）：HUMAN VIEW v2 — F11 technical amendment: candidate editing with minimum human friction 证据：`docs/HUMAN-VIEW-V2-F11-TECH.md`
- **HUMAN VIEW v2 — F12: the curation lane propose-apply — TECH AMENDMENT**（documentation）：HUMAN VIEW v2 — F12: the curation lane propose-apply — TECH AMENDMENT 证据：`docs/HUMAN-VIEW-V2-F12-TECH.md`
- **HUMAN VIEW v2 — F2.5 technical amendment: the mission tray, direct delivery, and the runner daemon**（documentation）：HUMAN VIEW v2 — F2.5 technical amendment: the mission tray, direct delivery, and the runner daemon 证据：`docs/HUMAN-VIEW-V2-F25-TECH.md`
- **HUMAN VIEW v2 — F30 technical amendment: the Universe landing surface, /api/universe , and the Landing**（documentation）：HUMAN VIEW v2 — F30 technical amendment: the Universe landing surface, /api/universe , and the Landing 证据：`docs/HUMAN-VIEW-V2-F30-UNIVERSE.md`
- **Human View v2 — PRD**（documentation）：The visual operating system for repos. Code → visible architecture → actionable blocks → missions for agents. 证据：`docs/HUMAN-VIEW-V2-PRD.md`
- **Human View v2 — the screen book**（documentation）：Every screen and sub-screen, as structural ASCII, in the ARTKIT v2.0.0 design language. 证据：`docs/HUMAN-VIEW-V2-SCREENS.md`
- **Human View v2 — UML atlas systems and subsystems**（documentation）：Human View v2 — UML atlas systems and subsystems 证据：`docs/HUMAN-VIEW-V2-UML.md`
- **IDE & Client Integration Matrix**（documentation）：This page tracks where major coding clients connect MCP servers, and how m1nd fits into each one. 证据：`docs/IDE-INTEGRATIONS.md`
- **M1ND-10 — G9 custody decision — 2026-07-21**（documentation）：M1ND-10 — G9 custody decision — 2026-07-21 证据：`docs/M1ND-10-G9-CUSTODY-DECISION-20260721.md`
- **M1ND-10 implementation handoff — 2026-07-19**（documentation）：M1ND-10 implementation handoff — 2026-07-19 证据：`docs/M1ND-10-HANDOFF-20260719.md`
- **M1nd 10 — PRD de convergência do organismo**（documentation）：M1nd 10 — PRD de convergência do organismo 证据：`docs/M1ND-10-PRD.md`
- **M1ND-10 public-path migration plan — 2026-07-19**（documentation）：M1ND-10 public-path migration plan — 2026-07-19 证据：`docs/M1ND-10-PUBLIC-PATH-MIGRATION-PLAN-20260719.md`
- **M1nd 10 — UML de sistemas, subsistemas e conexões**（documentation）：M1nd 10 — UML de sistemas, subsistemas e conexões 证据：`docs/M1ND-10-UML.md`
- **M1ND Guardian Method**（documentation）：Status: operational doctrine. This document is not a PRD/UML amendment, a gate receipt, or authority to commit, publish, install, activate, ratify, or land evidence. 证据：`docs/M1ND-GUARDIAN-METHOD.md`
- **MASSIF PRD**（documentation）：MASSIF is a Human View lens for reading a codebase as proof topography: a stable isometric maquette of blocks where each piece keeps its place across snapshots and carries the structural evidence state already computed by m1nd. 证据：`docs/MASSIF-PRD.md`
- **MASSIF Screens**（documentation）：These are structural wireframes only. They bind layout, hierarchy, information placement, labels, empty states, and operator copy. Final color, texture, materiality, and illustration language come from the parallel design calibration captured on 2026-07-06; this file does not bind visual finish. 证据：`docs/MASSIF-SCREENS.md`
- **MCP Host Refresh**（documentation）：Use this when the repo-local m1nd-mcp binary works, but a client such as Codex, Claude, Cursor, Windsurf, Cline, or another MCP host still exposes an older or partial tool surface. 证据：`docs/MCP-HOST-REFRESH.md`
- **The Medulla — PRD**（documentation）：The antifragile memory layer across brains · states, promotion, pull-only recall, delegation inheritance Two-Tier Slices 5–7, spec'd whole 证据：`docs/MEDULLA-PRD.md`
- **m1nd Next-Gen Agent-Native Runtime**（documentation）：⚠️ Grounding correction guardian, 2026-06-29 . This PRD was synthesized by a deep-research workflow 6 dimensions; 3 completed — concurrency / trust / context. Resolution was folded into Subsystem C as an honest non claim . The dedicated agent-MEMORY dimension failed validation and is an OPEN research gap — extend L1GHT for multi-agent / decay / cross-agent: TODO. . The research agents grounded against the STALE ~/m1nd worktree beta.8 . The current main is v1.1.0 ~/m1nd-night and DOES ship focus + sufficiency compute sufficiency , handle focus , … — verified present, 9 matches — so Subsystem C EXTENDS them, it does not create them ; the appendix line "sufficiency symbols not present" is an a… 证据：`docs/NEXTGEN-AGENT-PRD.md`
- **m1nd — THE ORGANISM FROM INSIDE + 360 — PRD**（documentation）：m1nd — THE ORGANISM FROM INSIDE + 360 — PRD 证据：`docs/ORGANISM-INSIDE-PRD.md`
- **THE m1nd ORGANISM — the constitution**（documentation）：THE m1nd ORGANISM — the constitution 证据：`docs/ORGANISM-PRD.md`
- **PATHOS — m1nd agent handoff**（documentation）：Read this first. Single source of truth for any chat / subagent / parallel session working on m1nd, so we don't re-derive state or contradict each other. Last checkpoint: 2026-07-30 checkpoint 38 — THE GATES REVEALED THEMSELVES AS ONE CHAIN, THE OWNER RAN HIS FIRST CEREMONY AND IT ANSWERED HONESTLY, AND THE PRODUCT'S BRAIN CAME HOME TO ITS OWN REPO . Checkpoint 38, condensed: fifteen PRs merged since checkpoint 37, measured 460– 466, 468– 472, 474, 477– 478 , seven more armed in flight 446, 467, 473, 475, 476, 479, 480 . 1 The ladder is a provenance chain, not a list. The owner ran G7 LIVE with his own hand — the program's FIRST owner-performed ceremony — and it answered NOT PROVEN: manifes… 证据：`docs/PATHOS.md`
- **The Soul — PRD**（documentation）：PATHOS native and verified · the agentic soul as a first-class m1nd type: anchored claims, two tissues, the curator at the gates 证据：`docs/SOUL-PRD.md`
- **PRD — transplant : o verbo de move por grafo**（documentation）：PRD — transplant : o verbo de move por grafo Design After do ciclo prove-first · 2026-07-20 · status: RATIFICADO PELO DONO "assinado", 2026-07-20 após verdict CHANGE aplicado 8 required changes . Decisões fixadas: D1 = symbol+paths canônico v1 node id condicionado a A1 · D2 = extents de ingest no Burst 0 · D3 = promover v1 com as 7 fronteiras, A7/A8 IMPOSTAS · D4 = nome transplant · D5 = opção b : transplant emite evento que envelhece receipts dos blocos tocados. Método: destilado dos artefatos das ondas com endereço de prova em toda frase normativa; confrontado pelas matrizes de condições do mundo e de interseção com o organismo. Nada aqui foi inventado na escrita — o que não estava provad… 证据：`docs/TRANSPLANT-PRD.md`
- **The Two-Tier Brain — PRD**（documentation）：Per-project brains + a shared medulla · the memory that travels with git official, proof-grown 证据：`docs/TWO-TIER-BRAIN-PRD.md`
- **The Organism — UML Atlas**（documentation）：Every system, subsystem and seam of m1nd as code-grounded UML — one master map, twenty-two system sheets, one consolidated gap ledger. 证据：`docs/UML-ORGANISM.md`
- **m1nd 360 — Self-Updating Manifesto + X-RAY Conformance + Seek Integration PRD & UML**（documentation）：m1nd 360 — Self-Updating Manifesto + X-RAY Conformance + Seek Integration PRD & UML 证据：`docs/X360-RUNTIME-PRD.md`
- **m1nd Deployment & Production Setup**（documentation）：m1nd can run as a per-host stdio process the IDE starts and stops it , but each such process loads its own graph — which can exceed 100MB — into RAM on every launch. For always-on use, run one persistent owner that keeps the graph resident, and point every IDE/agent at it through the native attach bridge m1nd-mcp --attach : a thin stdio↔HTTP client that loads no graph, builds no engines, and takes no lease. 证据：`docs/deployment.md`
- **m1nd Use Cases — Comprehensive Guide by Audience**（documentation）：m1nd Use Cases — Comprehensive Guide by Audience 证据：`docs/use-cases.md`
- **m1nd Benchmark Harness Spec**（documentation）：Manual benchmarking was enough to discover patterns, but it is too noisy for repeated product claims. 证据：`docs/benchmarks/BENCHMARK_HARNESS_SPEC_2026-03-24.md`
- **m1nd Next Patch Plan**（documentation）：Date: 2026-03-24 Repo: Source: benchmark findings from docs/BENCHMARK RESEARCH 2026-03-24.md 证据：`docs/benchmarks/BENCHMARK_PATCH_PLAN_2026-03-24.md`
- **m1nd Benchmark Research**（documentation）：Date: 2026-03-24 Repo: Status: working research log, not marketing copy 证据：`docs/benchmarks/BENCHMARK_RESEARCH_2026-03-24.md`
- **Bug-Hunt Benchmark Protocol**（documentation）：Date: 2026-05-14 Status: internal protocol, not a result claim 证据：`docs/benchmarks/BUG_HUNT_BENCHMARK_PROTOCOL_2026-05-14.md`
- **M1ND-10 G6 — the formal blind run, as one owner command**（documentation）：M1ND-10 G6 — the formal blind run, as one owner command 证据：`docs/benchmarks/G6-FORMAL-CEREMONY.md`
- **G7 LIVE — the owner's ceremony**（documentation）：Staged 2026-07-29 against main b96b191f m1nd 1.5.0 . Everything below is machine-checked except the two gestures that are the owner's by definition: the release build and the run itself. Nothing here starts, contacts, or mutates the installed owner on :1338 . 证据：`docs/benchmarks/G7-LIVE-CEREMONY.md`
- **M1ND-10 — G9 Secure Enclave custody ceremony — the owner's runbook**（documentation）：M1ND-10 — G9 Secure Enclave custody ceremony — the owner's runbook 证据：`docs/benchmarks/G9-CUSTODY-CEREMONY.md`
- **M1ND-10 — the owner's one-sitting completion script**（documentation）：M1ND-10 — the owner's one-sitting completion script 证据：`docs/benchmarks/M1ND-10-OWNER-SITTING.md`
- 其余 20 条证据见 `AI_CONTEXT_PACK.json` 或 `EVIDENCE_INDEX.json`。

## 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：`docs/README.md`, `docs/benchmarks/README.md`, `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：`docs/README.md`, `docs/benchmarks/README.md`, `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

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**：importance `high`
  - source_paths: README.md, m1nd-demo/package.json, m1nd-ui/package.json, package.json, skills/README.md
- **Lib 模块**：importance `high`
  - source_paths: m1nd-ui/src/lib/alerts.test.ts, m1nd-ui/src/lib/alerts.ts, m1nd-ui/src/lib/api.ts, m1nd-ui/src/lib/buildMap.test.ts, m1nd-ui/src/lib/buildMap.ts
- **Lib 模块**：importance `high`
  - source_paths: npm/lib/agent-cli.js, npm/lib/agent-schemas.js, npm/lib/cli.js, npm/lib/mcp-runtime-client.js
- **Lib 模块**：importance `high`
  - source_paths: m1nd-demo/src/lib/utils.ts

## Repo Inspection Evidence / 源码检查证据

- repo_clone_verified: true
- repo_inspection_verified: true
- repo_commit: `8e9f32552b3237e7ff0d7ca1deaf5dbf9eb25b10`
- inspected_files: `README.md`, `package.json`, `docs/AGENT-FIRST-DEMO.md`, `docs/AGENT-PACKS.md`, `docs/AGENT-TASKNOTES.md`, `docs/CASE-INTELLIGENCE-PRD.md`, `docs/FOCUS-RUNTIME-PRD.md`, `docs/GENESIS-INGEST-CONSUMERS-SPEC.md`, `docs/HOST-INTEGRATION-MATRIX.md`, `docs/HUMAN-LAYER-PRD.md`, `docs/HUMAN-VIEW-V2-F0-TECH.md`, `docs/HUMAN-VIEW-V2-F0C-TECH.md`, `docs/HUMAN-VIEW-V2-F11-TECH.md`, `docs/HUMAN-VIEW-V2-F12-TECH.md`, `docs/HUMAN-VIEW-V2-F25-TECH.md`, `docs/HUMAN-VIEW-V2-F30-UNIVERSE.md`, `docs/HUMAN-VIEW-V2-PRD.md`, `docs/HUMAN-VIEW-V2-SCREENS.md`, `docs/HUMAN-VIEW-V2-UML.md`, `docs/IDE-INTEGRATIONS.md`

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/maxkle1nz/m1nd | 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/maxkle1nz/m1nd | 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/maxkle1nz/m1nd | 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/maxkle1nz/m1nd | 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/maxkle1nz/m1nd | 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/maxkle1nz/m1nd | 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/maxkle1nz/m1nd | 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。