# hivemind - 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 hivemind 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：`harnesses/claude-code/skills/hivemind-goals/SKILL.md`, `harnesses/claude-code/skills/hivemind-graph/SKILL.md`, `harnesses/claude-code/skills/hivemind-memory/SKILL.md`, `harnesses/codex/skills/deeplake-memory/SKILL.md` 等 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：`harnesses/claude-code/skills/hivemind-goals/SKILL.md`, `harnesses/claude-code/skills/hivemind-graph/SKILL.md`, `harnesses/claude-code/skills/hivemind-memory/SKILL.md`, `harnesses/codex/skills/deeplake-memory/SKILL.md` 等 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`, `.claude-plugin/plugin.json`, `harnesses/claude-code/.claude-plugin/plugin.json`, `harnesses/codex/.codex-plugin/plugin.json` 等 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`, `harnesses/codex/INSTALL.md` Claim：`clm_0003` supported 0.86

## 怎么开始

- `npm i -g @deeplake/hivemind && hivemind install` 证据：`README.md` Claim：`clm_0006` supported 0.86
- `/plugin marketplace add activeloopai/hivemind` 证据：`README.md` Claim：`clm_0007` supported 0.86
- `/plugin install hivemind` 证据：`README.md` Claim：`clm_0008` supported 0.86
- `git clone https://github.com/activeloopai/hivemind.git ~/.codex/hivemind` 证据：`README.md` Claim：`clm_0009` supported 0.86
- `git clone https://github.com/activeloopai/hivemind.git` 证据：`README.md` Claim：`clm_0009` supported 0.86, `clm_0010` supported 0.86
- `npx @deeplake/hivemind@latest install` 证据：`harnesses/codex/INSTALL.md` Claim：`clm_0011` supported 0.86
- `npx @deeplake/hivemind@latest codex install` 证据：`harnesses/codex/INSTALL.md` Claim：`clm_0012` supported 0.86
- `npx @deeplake/hivemind@latest codex uninstall` 证据：`harnesses/codex/INSTALL.md` Claim：`clm_0013` supported 0.86

## 继续前判断卡

- **当前建议**：需要管理员/安全审批
- **为什么**：继续前可能涉及密钥、账号、外部服务或敏感上下文，建议先经过管理员或安全审批。

### 30 秒判断

- **现在怎么做**：需要管理员/安全审批
- **最小安全下一步**：先跑 Prompt Preview；若涉及凭证或企业环境，先审批再试装
- **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：`harnesses/claude-code/skills/hivemind-goals/SKILL.md`, `harnesses/claude-code/skills/hivemind-graph/SKILL.md`, `harnesses/claude-code/skills/hivemind-memory/SKILL.md`, `harnesses/codex/skills/deeplake-memory/SKILL.md` 等 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：`harnesses/claude-code/skills/hivemind-goals/SKILL.md`, `harnesses/claude-code/skills/hivemind-graph/SKILL.md`, `harnesses/claude-code/skills/hivemind-memory/SKILL.md`, `harnesses/codex/skills/deeplake-memory/SKILL.md` 等 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`, `.claude-plugin/plugin.json`, `harnesses/claude-code/.claude-plugin/plugin.json`, `harnesses/codex/.codex-plugin/plugin.json` 等 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`, `harnesses/codex/INSTALL.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`, `.claude-plugin/plugin.json`, `harnesses/claude-code/.claude-plugin/plugin.json`, `harnesses/claude-code/skills/hivemind-goals/SKILL.md` 等
- **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`, `.claude-plugin/plugin.json`, `harnesses/claude-code/.claude-plugin/plugin.json`, `harnesses/codex/.codex-plugin/plugin.json` 等
- **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`, `harnesses/codex/INSTALL.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`, `.claude-plugin/plugin.json`, `harnesses/claude-code/.claude-plugin/plugin.json`, `harnesses/claude-code/skills/hivemind-goals/SKILL.md` 等
- **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`, `.claude-plugin/plugin.json`, `README.md`, `harnesses/claude-code/.claude-plugin/plugin.json` 等
- **环境变量 / API Key**：项目入口文档明确出现 API key、token、secret 或账号凭证配置。 原因：如果真实安装需要凭证，应先使用测试凭证并经过权限/合规判断。 证据：`README.md`, `library/knowledge/private/auth/auth-architecture.md`, `library/knowledge/private/multi-tenant/org-workspace-model.md`, `library/knowledge/private/security/credential-storage.md` 等
- **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。）
- **不要使用真实生产凭证**：环境变量/API key 一旦进入宿主或工具链，可能产生账号和合规风险。（适用：出现 API、TOKEN、KEY、SECRET 等环境线索时。）
- **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。
- **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。
- **准备撤销测试 API key 或 token**：测试凭证泄露或误用时，可以快速止损。
- **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_0014` 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`, `.claude-plugin/plugin.json`, `harnesses/claude-code/.claude-plugin/plugin.json`, `harnesses/codex/.codex-plugin/plugin.json` 等 Claim：`clm_0015` 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`, `harnesses/codex/INSTALL.md` Claim：`clm_0016` 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：`harnesses/claude-code/skills/hivemind-goals/SKILL.md`, `harnesses/claude-code/skills/hivemind-graph/SKILL.md`, `harnesses/claude-code/skills/hivemind-memory/SKILL.md`, `harnesses/codex/skills/deeplake-memory/SKILL.md` 等 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`, `.claude-plugin/plugin.json`, `harnesses/claude-code/.claude-plugin/plugin.json`, `harnesses/codex/.codex-plugin/plugin.json` 等 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`, `harnesses/codex/INSTALL.md` Claim：`clm_0003` supported 0.86

### 上下文规模

- 文件总数：347
- 重要文件覆盖：40/347
- 证据索引条目：80
- 角色 / Skill 条目：11

### 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 hivemind 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 hivemind 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 hivemind 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 11 个角色 / Skill / 项目文档条目。

- **hivemind-goals**（skill）：Create, track and update team goals + KPIs via the Deeplake virtual filesystem at memory/goal/ and memory/kpi/. Use whenever the user mentions a goal, objective, KPI, target, milestone, or asks to track progress on something measurable. ALSO use when the user says "task", "todo", "work item", "remind me to", "fix X", or any actionable work item — the goal system replaced the legacy hivemind tasks CLI and now covers… 激活提示：当用户任务与“hivemind-goals”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：`harnesses/claude-code/skills/hivemind-goals/SKILL.md`
- **hivemind-graph**（skill）：Query the local code graph functions, classes, calls, imports through the Deeplake mount at memory/graph/. Use when the user asks structural questions about the codebase — "what calls X?", "what does Y import?", "where is Z defined?", "what's the architecture / which subsystems exist?", "what's the impact of changing this?". The graph is an AST-derived map of the repo, queried as files no build needed — it rebuilds… 激活提示：当用户任务与“hivemind-graph”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：`harnesses/claude-code/skills/hivemind-graph/SKILL.md`
- **hivemind-memory**（skill）：Global team and org memory powered by Activeloop. ALWAYS check BOTH built-in memory AND Hivemind memory when recalling information. 激活提示：当用户任务与“hivemind-memory”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：`harnesses/claude-code/skills/hivemind-memory/SKILL.md`
- **hivemind-memory**（skill）：Global team and org memory powered by Activeloop. ALWAYS check BOTH built-in memory AND Hivemind memory when recalling information. 激活提示：当用户任务与“hivemind-memory”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：`harnesses/codex/skills/deeplake-memory/SKILL.md`
- **hivemind-goals**（skill）：Create, track and update team goals + KPIs via the Deeplake virtual filesystem at memory/goal/ and memory/kpi/. Use whenever the user mentions a goal, objective, KPI, target, milestone, or asks to track progress on something measurable. ALSO use when the user says "task", "todo", "work item", "remind me to", "fix X", or any actionable work item — the goal system replaced the legacy hivemind tasks CLI and now covers… 激活提示：当用户任务与“hivemind-goals”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：`harnesses/codex/skills/hivemind-goals/SKILL.md`
- **hivemind-graph**（skill）：Query the local code graph functions, classes, calls, imports through the Deeplake mount at memory/graph/. Use when the user asks structural questions about the codebase — "what calls X?", "what does Y import?", "where is Z defined?", "what is the architecture / which subsystems exist?". The graph is an AST-derived map of the repo, queried as files no build needed — it rebuilds automatically . 激活提示：当用户任务与“hivemind-graph”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：`harnesses/codex/skills/hivemind-graph/SKILL.md`
- **hivemind-goals**（skill）：Create, track and update team goals + KPIs in Hivemind via the hivemind CLI. Use whenever the user mentions a goal, objective, KPI, target, milestone, or asks to track progress on something measurable. ALSO use when the user says "task", "todo", "work item", "remind me to", "fix X", or any actionable work item — the goal system replaced the legacy hivemind tasks CLI and now covers both objectives and tasks. 激活提示：当用户任务与“hivemind-goals”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：`harnesses/hermes/skills/hivemind-goals/SKILL.md`
- **hivemind-graph**（skill）：Query the local code graph functions, classes, calls, imports through the Deeplake mount at memory/graph/. Use when the user asks structural questions about the codebase — "what calls X?", "what does Y import?", "where is Z defined?", "what is the architecture / which subsystems exist?". The graph is an AST-derived map of the repo, queried as files no build needed — it rebuilds automatically . 激活提示：当用户任务与“hivemind-graph”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：`harnesses/hermes/skills/hivemind-graph/SKILL.md`
- **hivemind**（skill）：Global team and org memory powered by Activeloop. ALWAYS check BOTH built-in memory AND Hivemind memory when recalling information. 激活提示：当用户任务与“hivemind”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：`harnesses/openclaw/skills/SKILL.md`
- **hivemind-goals**（skill）：Create, track, and read team goals + KPIs via Hivemind from openclaw. Use whenever the user mentions a goal, objective, KPI, target, milestone, or asks to track progress on something measurable. ALSO use when the user says "task", "todo", "work item", "remind me to", "fix X", or any actionable work item — the goal system replaced the legacy hivemind tasks CLI and now covers both objectives and tasks. 激活提示：当用户任务与“hivemind-goals”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：`harnesses/openclaw/skills/hivemind-goals/SKILL.md`
- **hivemind-graph**（skill）：Query the local AST-derived code graph functions, classes, calls, imports for structural codebase questions — what calls X, what does Y import, where is Z defined, blast radius of a change. The graph rebuilds automatically after each agent turn; use hivemind graph search and hivemind graph neighborhood tools no manual build step . 激活提示：当用户任务与“hivemind-graph”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：`harnesses/openclaw/skills/hivemind-graph/SKILL.md`

## 证据索引

- 共索引 80 条证据。

- **Benchmarks**（documentation）：Auto-learning, cloud-backed shared brain for Claude Code • OpenClaw • Codex • Cursor • Hermes • pi • Claude Cowork Alpha agents. 证据：`README.md`
- **Library**（documentation）：Documentation root for this repository. Schema version: v2 . 证据：`library/README.md`
- **Hivemind**（documentation）：Cloud-backed shared memory for AI agents. Install once, memory persists across sessions, machines, and channels — and is shared with every teammate in your Deeplake org. 证据：`harnesses/openclaw/README.md`
- **Issues**（documentation）：Reactive bug and incident work IRDs , organized by lifecycle state. 证据：`library/issues/README.md`
- **Issues — Backlog**（documentation）：Tracked issues with a fix plan, not yet in active resolution. 证据：`library/issues/backlog/README.md`
- **Issues — Completed**（documentation）：Resolved IRD folders. Entire ird- - / folders land here when the GitHub issue closes and the fix is confirmed. Do not edit files here after landing. 证据：`library/issues/completed/README.md`
- **Issues — In Work**（documentation）：IRDs currently being resolved. Move from backlog/ → here when fix work starts, then completed/ when the GitHub issue closes. 证据：`library/issues/in-work/README.md`
- **Knowledge**（documentation）：Reference documentation for this repository, organized by audience. 证据：`library/knowledge/README.md`
- **Knowledge — Private**（documentation）：Internal documentation for engineers, product, and AI agents. 证据：`library/knowledge/private/README.md`
- **Knowledge — Public**（documentation）：Customer-facing documentation. Anything in this folder may eventually be published. 证据：`library/knowledge/public/README.md`
- **Notes**（documentation）：Human-only scratch space. Agents never read or write here. 证据：`library/notes/README.md`
- **Requirements**（documentation）：Product and feature work, organized by lifecycle state. 证据：`library/requirements/README.md`
- **Requirements — Backlog**（documentation）：Planned PRDs not yet in implementation. All new PRD folders are created here. 证据：`library/requirements/backlog/README.md`
- **Requirements — Completed**（documentation）：Shipped PRD folders. Entire prd- - / folders land here after the work ships and is confirmed in production. Do not edit files here after landing. 证据：`library/requirements/completed/README.md`
- **Requirements — In Work**（documentation）：PRDs currently being implemented. Folder location = lifecycle state. 证据：`library/requirements/in-work/README.md`
- **Requirements — Reports**（documentation）：Routine code-scan and audit reports not tied to any specific PRD. 证据：`library/requirements/reports/README.md`
- **Package**（package_manifest）：{ "name": "@deeplake/hivemind", "version": "0.7.118", "description": "Cloud-backed persistent shared memory for AI agents powered by Deeplake", "type": "module", "repository": { "type": "git", "url": "git+https://github.com/activeloopai/hivemind.git" }, "publishConfig": { "access": "public" }, "bin": { "hivemind": "bundle/cli.js" }, "files": "bundle", "harnesses/codex/bundle", "harnesses/codex/skills", "harnesses/cursor/bundle", "harnesses/hermes/bundle", "mcp/bundle", "harnesses/pi/extension-source", "harnesses/openclaw/dist", "harnesses/openclaw/skills", "harnesses/openclaw/openclaw.plugin.json", "harnesses/openclaw/package.json", ".claude-plugin", "scripts", "README.md", "LICENSE" , "scr… 证据：`package.json`
- **Package**（package_manifest）：{ "name": "hivemind-codex", "version": "0.7.118", "description": "Cloud-backed persistent shared memory for OpenAI Codex CLI powered by Deeplake", "type": "module" } 证据：`harnesses/codex/package.json`
- **Package**（package_manifest）：{ "name": "hivemind", "version": "0.7.118", "type": "module", "description": "Hivemind — cloud-backed persistent shared memory for AI agents, powered by DeepLake", "license": "Apache-2.0", "openclaw": { "extensions": "./dist/index.js" , "install": { "npmSpec": "@deeplake/hivemind", "minHostVersion": " =2026.3.22" }, "compat": { "pluginApi": " =1.0.0" }, "build": { "openclawVersion": "2026.3.22" } }, "files": "dist", "openclaw.plugin.json" , "scripts": { "build": "node esbuild.config.mjs" } } 证据：`harnesses/openclaw/package.json`
- **Installing Hivemind for Codex CLI**（documentation）：The fastest path installs hivemind into every AI coding assistant on your machine Claude Code, Codex, OpenClaw with one command: 证据：`harnesses/codex/INSTALL.md`
- **Architecture**（documentation）：Agent Mechanism Hooks/tools wired ------------------- ------------------------------------ ----------------------------------------------------------------------------------------- Claude Code Marketplace plugin SessionStart · UserPromptSubmit · PreToolUse · PostToolUse · Stop · SubagentStop · SessionEnd Codex ~/.codex/hooks.json SessionStart · UserPromptSubmit · PreToolUse Bash · PostToolUse · Stop OpenClaw Native extension at ~/.openclaw/extensions/hivemind/ agent end capture · before agent start recall · contracted tools hivemind search / read / index Cursor 1.7+ ~/.cursor/hooks.json sessionStart · beforeSubmitPrompt · postToolUse · afterAgentResponse · stop · sessionEnd Hermes Skill at… 证据：`docs/ARCHITECTURE.md`
- **Hivemind Goals**（skill_instruction）：Track goals and KPIs as Markdown files inside the Deeplake virtual filesystem. Each file is one row in a dedicated team-shared table — the path encodes the structural metadata, the file body holds the human-readable description. 证据：`harnesses/claude-code/skills/hivemind-goals/SKILL.md`
- **Hivemind Code Graph**（skill_instruction）：A deterministic, AST-derived map of the current repository — every function, class, method, interface, type, enum, const, and module, plus the edges between them calls , imports , extends , implements , method of . It is queried as synthesized files under the Deeplake mount; there are no real files on disk and no network call in the read path. 证据：`harnesses/claude-code/skills/hivemind-graph/SKILL.md`
- **Hivemind Memory**（skill_instruction）：You have TWO memory sources. ALWAYS check BOTH when the user asks you to recall, remember, or look up ANY information: 证据：`harnesses/claude-code/skills/hivemind-memory/SKILL.md`
- **Hivemind Memory**（skill_instruction）：You have persistent memory at ~/.deeplake/memory/ — global memory shared across all sessions, users, and agents in the org. 证据：`harnesses/codex/skills/deeplake-memory/SKILL.md`
- **Hivemind Goals**（skill_instruction）：Track goals and KPIs as Markdown files inside the Deeplake virtual filesystem. Each file is one row in a dedicated team-shared table — the path encodes the structural metadata, the file body holds the human-readable description. 证据：`harnesses/codex/skills/hivemind-goals/SKILL.md`
- **Hivemind Code Graph**（skill_instruction）：A deterministic, AST-derived map of the current repository — every function, class, method, interface, type, enum, const, and module, plus the edges between them calls , imports , extends , implements , method of . It is queried as synthesized files under the Deeplake mount; there are no real files on disk and no network call in the read path. 证据：`harnesses/codex/skills/hivemind-graph/SKILL.md`
- **Hivemind Goals — CLI only Hermes**（skill_instruction）：⚠️ CRITICAL: On this runtime Hermes , you MUST use the hivemind shell CLI for goals + KPIs. DO NOT use write file on ~/.deeplake/memory/goal/... paths — those writes go to the local filesystem and never reach the team-shared hivemind goals table. Other team members will NOT see them. 证据：`harnesses/hermes/skills/hivemind-goals/SKILL.md`
- **Hivemind Code Graph**（skill_instruction）：A deterministic, AST-derived map of the current repository — every function, class, method, interface, type, enum, const, and module, plus the edges between them calls , imports , extends , implements , method of . It is queried as synthesized files under the Deeplake mount; there are no real files on disk and no network call in the read path. 证据：`harnesses/hermes/skills/hivemind-graph/SKILL.md`
- **Hivemind Memory**（skill_instruction）：You have TWO memory sources. ALWAYS check BOTH when the user asks you to recall, remember, or look up ANY information: 证据：`harnesses/openclaw/skills/SKILL.md`
- **Hivemind Goals openclaw**（skill_instruction）：OpenClaw exposes purpose-built tools for goals + KPIs. Use them directly — do NOT try to write files via the host filesystem. 证据：`harnesses/openclaw/skills/hivemind-goals/SKILL.md`
- **Hivemind Code Graph OpenClaw**（skill_instruction）：A deterministic, AST-derived map of the current repository — every function, class, method, interface, type, enum, const, and module, plus the edges between them calls , imports , extends , implements , method of . 证据：`harnesses/openclaw/skills/hivemind-graph/SKILL.md`
- **Marketplace**（structured_config）：{ "name": "hivemind", "owner": { "name": "Activeloop", "email": "support@activeloop.ai" }, "metadata": { "description": "Cloud-backed persistent shared memory for AI agents powered by Deeplake", "version": "0.7.118" }, "plugins": { "name": "hivemind", "description": "Persistent shared memory powered by Deeplake — captures all session activity and provides cross-session, cross-agent memory search", "version": "0.7.118", "source": { "source": "git-subdir", "url": "https://github.com/activeloopai/hivemind.git", "path": "harnesses/claude-code", "sha": "c065c43b3965b5cdeee90d3627a2c771da1b20ba" }, "homepage": "https://github.com/activeloopai/hivemind" } } 证据：`.claude-plugin/marketplace.json`
- **Plugin**（structured_config）：{ "name": "hivemind", "description": "Cloud-backed persistent memory powered by Deeplake — read, write, and share memory across Claude Code sessions and agents", "version": "0.7.118", "author": { "name": "Activeloop", "url": "https://deeplake.ai" }, "homepage": "https://deeplake.ai", "repository": "https://github.com/activeloopai/hivemind", "license": "Apache-2.0", "keywords": "memory", "deeplake", "persistent-memory", "shared-memory", "agent-memory" } 证据：`.claude-plugin/plugin.json`
- **Plugin**（structured_config）：{ "name": "hivemind", "description": "Cloud-backed persistent memory powered by Deeplake — read, write, and share memory across Claude Code sessions and agents", "version": "0.7.118", "author": { "name": "Activeloop", "url": "https://deeplake.ai" }, "homepage": "https://deeplake.ai", "repository": "https://github.com/activeloopai/hivemind", "license": "Apache-2.0", "keywords": "memory", "deeplake", "persistent-memory", "shared-memory", "agent-memory" } 证据：`harnesses/claude-code/.claude-plugin/plugin.json`
- **Plugin**（structured_config）：{ "name": "hivemind", "version": "0.6.7", "description": "Cloud-backed persistent memory powered by Deeplake — read, write, and share memory across Codex sessions and agents", "skills": , "mcpServers": , "apps": , "interface": { "displayName": "Hivemind Memory", "shortDescription": "Persistent shared memory for AI agents powered by Deeplake", "longDescription": "Captures all session activity and provides cross-session, cross-agent memory search via ~/.deeplake/memory/. Memory is stored in Deeplake managed tables and shared across sessions, users, and agents in your organization.", "developerName": "Activeloop", "category": "productivity", "capabilities": "memory", "session-capture", "search… 证据：`harnesses/codex/.codex-plugin/plugin.json`
- **License**（source_file）：Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ 证据：`LICENSE`
- **Capture Tasks — turning conversation tangents into Hivemind goals**（documentation）：Capture Tasks — turning conversation tangents into Hivemind goals 证据：`docs/CAPTURE_TASKS.md`
- **Embeddings semantic search**（documentation）：Hivemind can run a local embedding daemon nomic-embed-text-v1.5, ~130 MB so that Grep over ~/.deeplake/memory/ uses hybrid semantic + lexical ranking instead of pure BM25. This is off by default — the daemon depends on @huggingface/transformers , which pulls onnxruntime-node and sharp ~600 MB total with native binaries . Shipping that with every agent install would 60× the install size for a feature most users don't need. 证据：`docs/EMBEDDINGS.md`
- **Skills skillify**（documentation）：Hivemind codifies recurring patterns from your team's recent sessions into reusable skills that propagate to every agent on your team — automatically. Same architecture as the wiki worker: an async background process that fires on Stop / SessionEnd, mines recent sessions in scope, asks Haiku whether the activity contains something worth keeping, and writes a SKILL.md if so. 证据：`docs/SKILLIFY.md`
- **Summaries**（documentation）：Hivemind doesn't just capture raw events — it also generates an AI-written wiki summary for each session and stores it in the memory table alongside its 768-dim summary embedding . The summary is what shows up when you Grep for past sessions or follow links from ~/.deeplake/memory/index.md . 证据：`docs/SUMMARIES.md`
- **Auth Architecture**（documentation）：Category: Auth Version: 1.0 Date: June 2026 Status: Active 证据：`library/knowledge/private/auth/auth-architecture.md`
- **Cursor Extension Architecture**（documentation）：Category: Frontend Version: 1.0 Date: June 2026 Status: Active 证据：`library/knowledge/private/frontend/cursor-extension-architecture.md`
- **CLI Command Architecture**（documentation）：Category: Operations Version: 1.0 Date: June 2026 Status: Active 证据：`library/knowledge/private/operations/cli-command-architecture.md`
- **Hooks**（structured_config）：{ "description": "Hivemind memory — captures all session activity and provides memory search via ~/.deeplake/memory", "hooks": { "SessionStart": { "hooks": { "type": "command", "command": "node \"${CLAUDE PLUGIN ROOT}/bundle/session-start.js\"", "timeout": 10 }, { "type": "command", "command": "node \"${CLAUDE PLUGIN ROOT}/bundle/session-notifications.js\"", "timeout": 8 }, { "type": "command", "command": "node \"${CLAUDE PLUGIN ROOT}/bundle/session-start-setup.js\"", "timeout": 120, "async": true } } , "UserPromptSubmit": { "hooks": { "type": "command", "command": "node \"${CLAUDE PLUGIN ROOT}/bundle/capture.js\"", "timeout": 10, "async": true }, { "type": "command", "command": "node \"${C… 证据：`harnesses/claude-code/hooks/hooks.json`
- **Hooks**（structured_config）：{ "hooks": { "SessionStart": { "matcher": "startup resume", "hooks": { "type": "command", "command": "node \"$CODEX PLUGIN ROOT/bundle/session-start.js\"", "timeout": 10 } } , "UserPromptSubmit": { "hooks": { "type": "command", "command": "node \"$CODEX PLUGIN ROOT/bundle/capture.js\"", "timeout": 10 } } , "PreToolUse": { "matcher": "Bash", "hooks": { "type": "command", "command": "node \"$CODEX PLUGIN ROOT/bundle/pre-tool-use.js\"", "timeout": 10 } } , "PostToolUse": { "hooks": { "type": "command", "command": "node \"$CODEX PLUGIN ROOT/bundle/capture.js\"", "timeout": 15 } } , "Stop": { "hooks": { "type": "command", "command": "node \"$CODEX PLUGIN ROOT/bundle/stop.js\"", "timeout": 30 },… 证据：`harnesses/codex/hooks/hooks.json`
- **Openclaw.Plugin**（structured_config）：{ "id": "hivemind", "name": "Hivemind", "description": "Cloud-backed shared memory powered by Deeplake — auto-capture and auto-recall across sessions, agents, and teammates", "skills": "./skills" , "contracts": { "tools": "hivemind search", "hivemind read", "hivemind index", "hivemind goal add", "hivemind kpi add", "hivemind graph search", "hivemind graph neighborhood" , "commands": "hivemind login", "hivemind capture", "hivemind whoami", "hivemind orgs", "hivemind switch org", "hivemind workspaces", "hivemind switch workspace", "hivemind setup", "hivemind version", "hivemind update", "hivemind autoupdate" , "memoryCorpusSupplements": true }, "uiHints": { "autoCapture": { "label": "Auto-Cap… 证据：`harnesses/openclaw/openclaw.plugin.json`
- **Embed Daemon**（source_file）：function socketPathFor uid, dir = DEFAULT SOCKET DIR function pidPathFor uid, dir = DEFAULT SOCKET DIR ⋮---- async function importFromCanonicalSharedDeps sharedDir = join homedir , ".hivemind", "embed-deps" async function importFromBareSpecifier function normalizeTransformersModule mod async function defaultImportTransformers canonical = importFromCanonicalSharedDeps, bare = importFromBareSpecifier ⋮---- async load addPrefix text, kind async embed text, kind = "document" async embedBatch texts, kind = "document" truncate vec ⋮---- function isDebug function log tag, msg ⋮---- appendFileSync LOG, ${ / @ PURE / new Date .toISOString } ${tag} ${msg} ⋮---- var log2 = m function getUid ⋮---- asyn… 证据：`embeddings/embed-daemon.js`
- **───────────────────────────────────────────────────────────────────────**（source_file）：set -u PASS=0 FAIL=0 SKIP=0 green { printf "\033 0;32m%s\033 0m" "$1"; } red { printf "\033 0;31m%s\033 0m" "$1"; } gray { printf "\033 0;90m%s\033 0m" "$1"; } yellow { printf "\033 0;33m%s\033 0m" "$1"; } ok { echo " $ green PASS $1"; PASS=$ PASS+1 ; } bad { echo " $ red FAIL $1${2:+ — $2}"; FAIL=$ FAIL+1 ; } skip { echo " $ gray SKIP $1${2:+ — $2}"; SKIP=$ SKIP+1 ; } section { echo; echo "$ yellow "▎ $1" "; } require jq { command -v jq /dev/null 2 &1 { echo "$ red "fatal:" jq not found on PATH install jq for this script to run " &2 exit 2 } } require jq ─────────────────────────────────────────────────────────────────────── Claude Code — marketplace plugin section "Claude Code" if -d "$HO… 证据：`scripts/verify-install.sh`
- **Config**（source_file）：import { readFileSync, existsSync } from "node:fs"; import { join } from "node:path"; import { homedir, userInfo } from "node:os"; ⋮---- export interface Config { token: string; orgId: string; orgName: string; userName: string; workspaceId: string; apiUrl: string; tableName: string; sessionsTableName: string; skillsTableName: string; rulesTableName: string; goalsTableName: string; kpisTableName: string; codebaseTableName: string; memoryPath: string; } ⋮---- interface Credentials { token: string; orgId: string; orgName?: string; userName?: string; workspaceId?: string; apiUrl?: string; } ⋮---- export function loadConfig : Config null 证据：`src/config.ts`
- **Deeplake Api**（source_file）：import { randomUUID } from "node:crypto"; import { log as log } from "./utils/debug.js"; import { sqlStr, sqlIdent } from "./utils/sql.js"; import { SUMMARY EMBEDDING COL } from "./embeddings/columns.js"; import { deeplakeClientHeader } from "./utils/client-header.js"; import { CODEBASE COLUMNS, MEMORY COLUMNS, SESSIONS COLUMNS, SKILLS COLUMNS, RULES COLUMNS, GOALS COLUMNS, KPIS COLUMNS, buildCreateTableSql, healMissingColumns, } from "./deeplake-schema.js"; import { enqueueNotification } from "./notifications/queue.js"; import { loadCredentials } from "./commands/auth-creds.js"; ⋮---- type IndexMarkerStore = typeof import "./index-marker-store.js" ; ⋮---- function getIndexMarkerStore : Pro… 证据：`src/deeplake-api.ts`
- **Deeplake Schema**（source_file）：import { sqlIdent, sqlStr } from "./utils/sql.js"; ⋮---- export interface ColumnDef { name: string; sql: string; } ⋮---- function validateSchema label: string, cols: readonly ColumnDef : void ⋮---- export function buildCreateTableSql tableName: string, cols: readonly ColumnDef : string ⋮---- function buildIntrospectionSql tableName: string, workspaceId: string : string ⋮---- export type QueryFn = sql: string = Promise ; ⋮---- export interface HealResult { missing: string ; altered: string ; } ⋮---- export async function healMissingColumns args: { query: QueryFn; tableName: string; workspaceId: string; columns: readonly ColumnDef ; log?: msg: string ⋮---- export function isMissingTableError… 证据：`src/deeplake-schema.ts`
- **Index Marker Store**（source_file）：import { existsSync, mkdirSync, readFileSync, writeFileSync } from "node:fs"; import { join } from "node:path"; import { tmpdir } from "node:os"; ⋮---- export function getIndexMarkerDir : string ⋮---- export function buildIndexMarkerPath workspaceId: string, orgId: string, table: string, suffix: string : string ⋮---- export function hasFreshIndexMarker markerPath: string : boolean ⋮---- export function writeIndexMarker markerPath: string : void 证据：`src/index-marker-store.ts`
- **User Config**（source_file）：import { existsSync, mkdirSync, readFileSync, renameSync, writeFileSync } from "node:fs"; import { homedir } from "node:os"; import { dirname, join } from "node:path"; ⋮---- export interface UserConfig { embeddings?: { enabled?: boolean; }; } ⋮---- let configPath: ⋮---- export function readUserConfig : UserConfig ⋮---- export function writeUserConfig patch: Partial : UserConfig ⋮---- export function getEmbeddingsEnabled : boolean ⋮---- function migrationValueFromEnv : boolean ⋮---- export function setEmbeddingsEnabled enabled: boolean : void ⋮---- function isPlainObject value: unknown : value is Record ⋮---- function deepMerge base: UserConfig, patch: Partial : UserConfig ⋮---- export funct… 证据：`src/user-config.ts`
- **Vitest.Config**（source_file）：import { defineConfig } from "vitest/config"; 证据：`vitest.config.ts`
- **Install**（source_file）：set -e exec npx -y @deeplake/hivemind@latest codex install 证据：`harnesses/codex/install.sh`
- **Index**（source_file）：function definePluginEntry entry: T : T ⋮---- type SetupConfigModule = typeof import "./setup-config.js" ; function loadSetupConfig : Promise ⋮---- import { requestDeviceCode, pollForToken, listOrgs, switchOrg, listWorkspaces, switchWorkspace, healDriftedOrgToken } from "../../../src/commands/auth.js"; import { DeeplakeApi } from "../../../src/deeplake-api.js"; ⋮---- type CredsModule = typeof import "../../../src/commands/auth-creds.js" ; type ConfigModule = typeof import "../../../src/config.js" ; ⋮---- function loadCredsModule : Promise function loadConfigModule : Promise async function loadCredentials async function saveCredentials creds: Awaited : Promise async function loadConfig impor… 证据：`harnesses/openclaw/src/index.ts`
- **Setup Config**（source_file）：import { existsSync, readFileSync, writeFileSync, renameSync } from "node:fs"; import { homedir } from "node:os"; import { join } from "node:path"; ⋮---- export function getOpenclawConfigPath : string ⋮---- export function isAllowlistCoveringHivemind alsoAllow: unknown : boolean ⋮---- export function isPluginsAllowMissingHivemind allow: unknown : boolean ⋮---- export type AllowlistDelta = { pluginsAllow: boolean; toolsAlsoAllow: boolean; }; ⋮---- export type SetupResult = { status: "already-set"; configPath: string } { status: "added"; configPath: string; backupPath: string; delta: AllowlistDelta } { status: "error"; configPath: string; error: string }; ⋮---- export function ensureHivemindA… 证据：`harnesses/openclaw/src/setup-config.ts`
- **Index**（source_file）：import { installClaude, uninstallClaude } from "./install-claude.js"; import { installCodex, uninstallCodex } from "./install-codex.js"; import { installOpenclaw, uninstallOpenclaw } from "./install-openclaw.js"; import { installCursor, uninstallCursor } from "./install-cursor.js"; import { installHermes, uninstallHermes } from "./install-hermes.js"; import { installCowork, uninstallCowork } from "./install-cowork.js"; import { installPi, uninstallPi } from "./install-pi.js"; import { disableEmbeddings, enableEmbeddings, installEmbeddings, statusEmbeddings, uninstallEmbeddings, } from "./embeddings.js"; import { ensureLoggedIn, isLoggedIn, loginWithProvidedToken, maybeShowOrgChoice } from "… 证据：`src/cli/index.ts`
- **Daemon**（source_file）：import { createServer, type Server, type Socket } from "node:net"; import { unlinkSync, writeFileSync, existsSync, mkdirSync, chmodSync } from "node:fs"; import { NomicEmbedder } from "./nomic.js"; import { DEFAULT IDLE TIMEOUT MS, PROTOCOL VERSION, pidPathFor, socketPathFor, type DaemonRequest, type DaemonResponse, type EmbedRequest, type HelloRequest, type PingRequest, } from "./protocol.js"; import { log as log } from "../utils/debug.js"; ⋮---- const log = m: string ⋮---- function getUid : string ⋮---- export interface DaemonOptions { socketDir?: string; idleTimeoutMs?: number; dims?: number; dtype?: string; repo?: string; daemonPath?: string; } ⋮---- export class EmbedDaemon ⋮---- const… 证据：`src/embeddings/daemon.ts`
- 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：`README.md`, `library/README.md`, `harnesses/openclaw/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：`README.md`, `library/README.md`, `harnesses/openclaw/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。

- **项目概述与系统架构**：importance `high`
  - source_paths: README.md, docs/ARCHITECTURE.md, package.json, src/cli/index.ts, src/deeplake-api.ts
- **核心功能与数据流（捕获、回忆、技能化、代码图谱）**：importance `high`
  - source_paths: src/hooks/capture.ts, src/hooks/recall.ts, src/hooks/wiki-worker.ts, src/skillify/skillify-worker.ts, src/graph/graph-command.ts
- **多 Agent 集成与 Harness 适配层**：importance `high`
  - source_paths: harnesses/claude-code/hooks/hooks.json, harnesses/codex/hooks/hooks.json, harnesses/codex/install.sh, harnesses/codex/INSTALL.md, harnesses/openclaw/openclaw.plugin.json
- **运维、故障模式与社区热点问题**：importance `high`
  - source_paths: src/hooks/recall.ts, src/embeddings/standalone-embed-client.ts, src/embeddings/daemon.ts, src/embeddings/self-heal.ts, src/index-marker-store.ts

## Repo Inspection Evidence / 源码检查证据

- repo_clone_verified: true
- repo_inspection_verified: true
- repo_commit: `64ab629cc9261ee8511eb327f2dfb05a3fee4b10`
- inspected_files: `README.md`, `package.json`, `docs/ARCHITECTURE.md`, `docs/CAPTURE_TASKS.md`, `docs/EMBEDDINGS.md`, `docs/SKILLIFY.md`, `docs/SUMMARIES.md`, `src/cli/agents-md.ts`, `src/cli/auth.ts`, `src/cli/embeddings.ts`, `src/cli/index.ts`, `src/cli/install-claude.ts`, `src/cli/install-codex.ts`, `src/cli/install-cowork.ts`, `src/cli/install-cursor.ts`, `src/cli/install-hermes.ts`, `src/cli/install-mcp-shared.ts`, `src/cli/install-openclaw.ts`, `src/cli/install-pi.ts`, `src/cli/install-scan.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: 失败模式：installation: Recall UserPromptSubmit hook hangs (2s timeout) when embed-daemon.js launcher is missing from...

- Trigger: Developers should check this installation risk before relying on the project: Recall UserPromptSubmit hook hangs (2s timeout) when embed-daemon.js launcher is missing from embed-deps despite embeddings enabled
- Host AI rule: Before packaging this project, run the relevant install/config/quickstart check for: Recall UserPromptSubmit hook hangs (2s timeout) when embed-daemon.js launcher is missing from embed-deps despite embeddings enabled. Context: Observed when using node, macos
- Why it matters: Developers may fail before the first successful local run: Recall UserPromptSubmit hook hangs (2s timeout) when embed-daemon.js launcher is missing from embed-deps despite embeddings enabled
- Evidence: failure_mode_cluster:github_issue | https://github.com/activeloopai/hivemind/issues/296 | Recall UserPromptSubmit hook hangs (2s timeout) when embed-daemon.js launcher is missing from embed-deps despite embeddings enabled
- 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: 失败模式：installation: v0.7.116 — fix(cli): don't crash when optional tree-sitter addon is absent (install P0)

- Trigger: Developers should check this installation risk before relying on the project: v0.7.116 — fix(cli): don't crash when optional tree-sitter addon is absent (install P0)
- Host AI rule: Before packaging this project, run the relevant install/config/quickstart check for: v0.7.116 — fix(cli): don't crash when optional tree-sitter addon is absent (install P0). Context: Observed during installation or first-run setup.
- Why it matters: Upgrade or migration may change expected behavior: v0.7.116 — fix(cli): don't crash when optional tree-sitter addon is absent (install P0)
- Evidence: failure_mode_cluster:github_release | https://github.com/activeloopai/hivemind/releases/tag/v0.7.116 | v0.7.116 — fix(cli): don't crash when optional tree-sitter addon is absent (install P0)
- 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: 来源证据：Recall UserPromptSubmit hook hangs (2s timeout) when embed-daemon.js launcher is missing from embed-deps despite embedd…

- 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：Recall UserPromptSubmit hook hangs (2s timeout) when embed-daemon.js launcher is missing from embed-deps despite embeddings enabled
- Why it matters: 可能阻塞安装或首次运行。
- Evidence: community_evidence:github | https://github.com/activeloopai/hivemind/issues/296 | 来源讨论提到 node 相关条件，需在安装/试用前复核。
- 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 4: 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/activeloopai/hivemind | host_targets=claude, openclaw, claude_code, cursor
- 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: 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/activeloopai/hivemind | 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 6: 失败模式：runtime: session-start: CREATE INDEX IF NOT EXISTS times out at 10s on busy sessions tables

- Trigger: Developers should check this runtime risk before relying on the project: session-start: CREATE INDEX IF NOT EXISTS times out at 10s on busy sessions tables
- Host AI rule: Before packaging this project, run the relevant install/config/quickstart check for: session-start: CREATE INDEX IF NOT EXISTS times out at 10s on busy sessions tables. Context: Source discussion did not expose a precise runtime context.
- Why it matters: Developers may hit a documented source-backed failure mode: session-start: CREATE INDEX IF NOT EXISTS times out at 10s on busy sessions tables
- Evidence: failure_mode_cluster:github_issue | https://github.com/activeloopai/hivemind/issues/89 | session-start: CREATE INDEX IF NOT EXISTS times out at 10s on busy sessions tables
- 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: 失败模式：runtime: v0.7.117 — fix: stop wiki summary worker crash on long sessions

- Trigger: Developers should check this runtime risk before relying on the project: v0.7.117 — fix: stop wiki summary worker crash on long sessions
- Host AI rule: Before packaging this project, run the relevant install/config/quickstart check for: v0.7.117 — fix: stop wiki summary worker crash on long sessions. Context: Source discussion did not expose a precise runtime context.
- Why it matters: Upgrade or migration may change expected behavior: v0.7.117 — fix: stop wiki summary worker crash on long sessions
- Evidence: failure_mode_cluster:github_release | https://github.com/activeloopai/hivemind/releases/tag/v0.7.117 | v0.7.117 — fix: stop wiki summary worker crash on long sessions
- 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 8: 失败模式：runtime: v0.7.118 — fix: stop wiki summary worker crash on long sessions

- Trigger: Developers should check this runtime risk before relying on the project: v0.7.118 — fix: stop wiki summary worker crash on long sessions
- Host AI rule: Before packaging this project, run the relevant install/config/quickstart check for: v0.7.118 — fix: stop wiki summary worker crash on long sessions. Context: Source discussion did not expose a precise runtime context.
- Why it matters: Upgrade or migration may change expected behavior: v0.7.118 — fix: stop wiki summary worker crash on long sessions
- Evidence: failure_mode_cluster:github_release | https://github.com/activeloopai/hivemind/releases/tag/v0.7.118 | v0.7.118 — fix: stop wiki summary worker crash on long sessions
- 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 9: 来源证据：session-start: CREATE INDEX IF NOT EXISTS times out at 10s on busy sessions tables

- Trigger: GitHub 社区证据显示该项目存在一个运行相关的待验证问题：session-start: CREATE INDEX IF NOT EXISTS times out at 10s on busy sessions tables
- 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 | https://github.com/activeloopai/hivemind/issues/89 | 来源类型 github_issue 暴露的待验证使用条件。
- 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 10: 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/activeloopai/hivemind | 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。