# codelayer - 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 codelayer 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

- `npx humanlayer join-waitlist --email ...` 证据：`README.md` Claim：`clm_0003` supported 0.86

## 继续前判断卡

- **当前建议**：先做角色匹配试用
- **为什么**：这个项目更像角色库，核心风险是选错角色或把角色文案当执行能力；先用 Prompt Preview 试角色匹配，再决定是否沙盒导入。

### 30 秒判断

- **现在怎么做**：先做角色匹配试用
- **最小安全下一步**：先用 Prompt Preview 试角色匹配；满意后再隔离导入
- **先别相信**：角色质量和任务匹配不能直接相信。
- **继续会触碰**：角色选择偏差、命令执行、宿主 AI 配置

### 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

- **角色质量和任务匹配不能直接相信。**（unverified）：角色库证明有很多角色，不证明每个角色都适合你的具体任务，也不证明角色能产生高质量结果。
- **不能把角色文案当成真实执行能力。**（unverified）：安装前只能判断角色描述和任务画像是否匹配，不能证明它能在宿主 AI 里完成任务。
- **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.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。

### 继续会触碰什么

- **角色选择偏差**：用户对任务应该由哪个专家角色处理的判断。 原因：选错角色会让 AI 从错误专业视角回答，浪费时间或误导决策。
- **命令执行**：包管理器、网络下载、本地插件目录、项目配置或用户主目录。 原因：运行第一条命令就可能产生环境改动；必须先判断是否值得跑。 证据：`README.md`
- **宿主 AI 配置**：Claude/Codex/Cursor/Gemini/OpenCode 等宿主的 plugin、Skill 或规则加载配置。 原因：宿主配置会改变 AI 后续工作方式，可能和用户已有规则冲突。 证据：`CLAUDE.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**：先用交互式试用验证任务画像和角色匹配，不要先导入整套角色库。（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。
- **保留原始角色选择记录**：如果输出偏题，可以回到任务画像阶段重新选择角色，而不是继续沿着错误角色推进。
- **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_0004` 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_0005` 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：729
- 重要文件覆盖：40/729
- 证据索引条目：80
- 角色 / Skill 条目：68

### 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 codelayer 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 codelayer 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 codelayer 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 68 个角色 / Skill / 项目文档条目。

- **Mintlify Starter Kit**（project_doc）：Click on Use this template to copy the Mintlify starter kit. The starter kit contains examples including 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/readme.md`
- **CLAUDE.md**（project_doc）：This file provides guidance to Claude Code claude.ai/code when working with code in this repository. 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：`CLAUDE.md`
- **The best way to get Coding Agents to solve hard problems in complex codebases**（project_doc）：! Wordmark Logo of HumanLayer ./docs/images/wordmark-light.svg 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`
- **Claude Code Go SDK Experimental**（project_doc）：A Go SDK for programmatically interacting with Claude Code Anthropic's AI coding assistant . 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：`claudecode-go/README.md`
- **Go style guidelines**（project_doc）：This is the humanlayer Daemon HLD that powers the WUI humanlayer-wui 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：`hld/CLAUDE.md`
- **HumanLayer Daemon hld**（project_doc）：The HumanLayer Daemon hld provides a REST API and JSON-RPC interface for managing Claude Code sessions, approvals, and real-time event streaming. 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：`hld/README.md`
- **HumanLayer CLI**（project_doc）：HumanLayer, but on your command-line. 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：`hlyr/README.md`
- **Tips and Tricks**（project_doc）：This is the HumanLayer Web UI WUI - a desktop application for managing AI agent approvals and sessions. 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：`humanlayer-wui/CLAUDE.md`
- **humanlayer-wui**（project_doc）：Web/desktop UI for the HumanLayer daemon hld built with Tauri and React. 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：`humanlayer-wui/README.md`
- **APIs**（project_doc）：Default to using Bun instead of Node.js. 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：`scripts/CLAUDE.md`
- **scripts**（project_doc）：This project was created using bun init in bun v1.2.23. Bun https://bun.com is a fast all-in-one JavaScript runtime. 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：`scripts/README.md`
- **HOW TO ITERATE ON GITHUB ACTIONS**（project_doc）：1. Select a branch to work on 2. Add branch to workflow trigger : 3. Make one targeted change e.g., cache one thing like Go dependencies 4. Commit and push to the branch ON UPSTREAM git push -u upstream your-debug-branch 5. Check logs with gh : 6. Repeat until working as expected 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：`.github/workflows/CLAUDE.md`
- **APIs**（project_doc）：Default to using Bun instead of Node.js. 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：`apps/daemon/CLAUDE.md`
- **daemon**（project_doc）：This project was created using bun init in bun v1.2.23. Bun https://bun.com is a fast all-in-one JavaScript runtime. 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：`apps/daemon/README.md`
- **APIs**（project_doc）：Default to using Bun instead of Node.js. 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：`apps/react/CLAUDE.md`
- **bun-react-tailwind-shadcn-template**（project_doc）：This project was created using bun init in bun v1.2.23. Bun https://bun.com is a fast all-in-one JavaScript runtime. 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：`apps/react/README.md`
- **Linear CLI**（project_doc）：A command-line interface for interacting with Linear issue tracking. 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：`hack/linear/README.md`
- **HumanLayer Daemon TypeScript SDK**（project_doc）：This SDK provides a TypeScript/JavaScript client for the HumanLayer Daemon HLD REST API. 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：`hld/sdk/typescript/README.md`
- **WUI Demo Store System**（project_doc）：A comprehensive demo store system for creating synthetic product shots and marketing animations using Zustand's slice pattern architecture. 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：`humanlayer-wui/src/stores/demo/README.md`
- **APIs**（project_doc）：Default to using Bun instead of Node.js. 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：`packages/contracts/CLAUDE.md`
- **@codelayer/protocol**（project_doc）：This project was created using bun init in bun v1.2.23. Bun https://bun.com is a fast all-in-one JavaScript runtime. 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：`packages/contracts/README.md`
- **APIs**（project_doc）：Default to using Bun instead of Node.js. 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：`packages/database/CLAUDE.md`
- **database**（project_doc）：This project was created using bun init in bun v1.2.23. Bun https://bun.com is a fast all-in-one JavaScript runtime. 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：`packages/database/README.md`
- **Contributing to HumanLayer**（project_doc）：If you're looking to contribute, please: 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`
- **Documentation System**（project_doc）：- Version tags follow semver vX.Y.Z - Features added in main branch - Examples updated alongside feature development - Changelog maintained for each version - Both Python and TypeScript packages versioned together - Generate release notes using git commands: - Use git diff v0.5.11..v0.6.0 to see file changes between versions - Always query changes from git before updating CHANGELOG.md - Changelog priorities: - Docum… 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/docs.knowledge.md`
- **Case Study: Accelerating Technical Debt Resolution with AI Coding Agents**（project_doc）：Case Study: Accelerating Technical Debt Resolution with AI Coding Agents 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/case-studies/healthcare-case-study.md`
- **API Reference**（project_doc）：Fetch and manage approval requests. 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：`humanlayer-wui/docs/API.md`
- **Architecture Overview**（project_doc）：mermaid graph LR subgraph "Frontend TypeScript " RC React Component RH React Hooks DC Daemon Client end 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：`humanlayer-wui/docs/ARCHITECTURE.md`
- **Developer Guide**（project_doc）：This guide helps you build UI components that interact with the HumanLayer daemon. 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：`humanlayer-wui/docs/DEVELOPER_GUIDE.md`
- **Hotkey Scope System Documentation**（project_doc）：The application uses a hierarchical hotkey scope system built on top of react-hotkeys-hook to properly isolate keyboard shortcuts and prevent conflicts between different UI contexts. 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：`humanlayer-wui/docs/HOTKEYS.md`
- **What problem s was I solving?**（project_doc）：What user-facing changes did I ship? 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：`.github/PULL_REQUEST_TEMPLATE.md`
- **Development Guide**（project_doc）：This guide covers development workflows and tools for the HumanLayer repository. 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：`DEVELOPMENT.md`
- **HumanLayer Daemon HLD Protocol Documentation**（project_doc）：HumanLayer Daemon HLD Protocol Documentation 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：`hld/PROTOCOL.md`
- **HumanLayer Daemon HLD - TODO**（project_doc）：Goal : Reduce N+1 problem for TUI message count display Current issue : TUI needs separate GetConversation call per session to count messages Proposed solution : Add bulk endpoint that returns conversation metadata message count, last message, etc. for multiple sessions Alternative : Extend ListSessions to include conversation metadata Performance impact : Would significantly improve TUI session list load times File… 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：`hld/TODO.md`
- **Changelog**（project_doc）：All notable changes to the HumanLayer CLI hlyr will be documented in this file. 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：`hlyr/CHANGELOG.md`
- **HumanLayer Thoughts Management System**（project_doc）：HumanLayer Thoughts Management System 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：`hlyr/THOUGHTS.md`
- **macOS Release Workflow Usage Guide**（project_doc）：Testing the Workflow Before Merging 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：`humanlayer-wui/RELEASE.md`
- **Problems**（project_doc）：- 'c' on session table doesn't launch the session creator - selecting "create new session" from the cmd+k launch takes me to a blank screen - search view still doesn't work - the max height should be 80% of the screen height, and only display as many items as can fit - selecting a session from the search view should navigate to the session details 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：`humanlayer-wui/problems.md`
- **HumanLayer SDK Documentation**（project_doc）：🚧 Note : This documentation covers the HumanLayer SDK which is being superseded by CodeLayer. For the latest IDE experience, see the main README. 🚧 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：`humanlayer.md`
- **Manual Test Plan for Slash Commands Fix**（project_doc）：Manual Test Plan for Slash Commands Fix 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：`test-slash-commands.md`
- **codebase-analyzer**（project_doc）：Analyzes codebase implementation details. Call the codebase-analyzer agent when you need to find detailed information about specific components. As always, the more detailed your request prompt, the better! : 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：`.claude/agents/codebase-analyzer.md`
- **codebase-locator**（project_doc）：Locates files, directories, and components relevant to a feature or task. Call codebase-locator with human language prompt describing what you're looking for. Basically a "Super Grep/Glob/LS tool" — Use it if you find yourself desiring to use one of these tools more than once. 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：`.claude/agents/codebase-locator.md`
- **codebase-pattern-finder**（project_doc）：codebase-pattern-finder is a useful subagent type for finding similar implementations, usage examples, or existing patterns that can be modeled after. It will give you concrete code examples based on what you're looking for! It's sorta like codebase-locator, but it will not only tell you the location of files, it will also give you code details! 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：`.claude/agents/codebase-pattern-finder.md`
- **thoughts-analyzer**（project_doc）：The research equivalent of codebase-analyzer. Use this subagent type when wanting to deep dive on a research topic. Not commonly needed otherwise. 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：`.claude/agents/thoughts-analyzer.md`
- **thoughts-locator**（project_doc）：Discovers relevant documents in thoughts/ directory We use this for all sorts of metadata storage! . This is really only relevant/needed when you're in a reseaching mood and need to figure out if we have random thoughts written down that are relevant to your current research task. Based on the name, I imagine you can guess this is the thoughts equivilent of codebase-locator 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：`.claude/agents/thoughts-locator.md`
- **web-search-researcher**（project_doc）：Do you find yourself desiring information that you don't quite feel well-trained confident on? Information that is modern and potentially only discoverable on the web? Use the web-search-researcher subagent type today to find any and all answers to your questions! It will research deeply to figure out and attempt to answer your questions! If you aren't immediately satisfied you can get your money back! Not really -… 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：`.claude/agents/web-search-researcher.md`
- **Commit Changes**（project_doc）：Create git commits for session changes with clear, atomic messages 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：`.claude/commands/ci_commit.md`
- **Generate PR Description**（project_doc）：Generate comprehensive PR descriptions following repository templates 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：`.claude/commands/ci_describe_pr.md`
- **Commit Changes**（project_doc）：Create git commits with user approval and no Claude attribution 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：`.claude/commands/commit.md`
- **Create Handoff**（project_doc）：Create handoff document for transferring work to another session 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：`.claude/commands/create_handoff.md`
- **Implementation Plan**（project_doc）：Create detailed implementation plans through interactive research and iteration 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：`.claude/commands/create_plan.md`
- **Implementation Plan**（project_doc）：Create detailed implementation plans with thorough research and iteration 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：`.claude/commands/create_plan_generic.md`
- **Implementation Plan**（project_doc）：Create implementation plans with thorough research no thoughts directory 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：`.claude/commands/create_plan_nt.md`
- **Create Worktree**（project_doc）：Create worktree and launch implementation session for a plan 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：`.claude/commands/create_worktree.md`
- **Debug**（project_doc）：Debug issues by investigating logs, database state, and git history 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：`.claude/commands/debug.md`
- **Generate PR Description**（project_doc）：Generate comprehensive PR descriptions following repository templates 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：`.claude/commands/describe_pr.md`
- **Generate PR Description**（project_doc）：Generate comprehensive PR descriptions following repository templates 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：`.claude/commands/describe_pr_nt.md`
- **Founder Mode**（project_doc）：Create Linear ticket and PR for experimental features after implementation 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：`.claude/commands/founder_mode.md`
- **Implement Plan**（project_doc）：Implement technical plans from thoughts/shared/plans with verification 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：`.claude/commands/implement_plan.md`
- **Iterate Implementation Plan**（project_doc）：Iterate on existing implementation plans with thorough research and updates 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：`.claude/commands/iterate_plan.md`
- **Iterate Implementation Plan**（project_doc）：Iterate on existing implementation plans with thorough research and updates 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：`.claude/commands/iterate_plan_nt.md`
- **Linear - Ticket Management**（project_doc）：Manage Linear tickets - create, update, comment, and follow workflow patterns 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：`.claude/commands/linear.md`
- **Local Review**（project_doc）：Set up worktree for reviewing colleague's branch 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：`.claude/commands/local_review.md`
- **Oneshot**（project_doc）：Research ticket and launch planning session 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：`.claude/commands/oneshot.md`
- **Oneshot Plan**（project_doc）：Execute ralph plan and implementation for a ticket 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：`.claude/commands/oneshot_plan.md`
- **PART I - IF A TICKET IS MENTIONED**（project_doc）：Implement highest priority small Linear ticket with worktree setup 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：`.claude/commands/ralph_impl.md`
- **PART I - IF A TICKET IS MENTIONED**（project_doc）：Create implementation plan for highest priority Linear ticket ready for spec 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：`.claude/commands/ralph_plan.md`
- **PART I - IF A LINEAR TICKET IS MENTIONED**（project_doc）：Research highest priority Linear ticket needing investigation 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：`.claude/commands/ralph_research.md`

## 证据索引

- 共索引 80 条证据。

- **Mintlify Starter Kit**（documentation）：Click on Use this template to copy the Mintlify starter kit. The starter kit contains examples including 证据：`docs/readme.md`
- **CLAUDE.md**（documentation）：This file provides guidance to Claude Code claude.ai/code when working with code in this repository. 证据：`CLAUDE.md`
- **The best way to get Coding Agents to solve hard problems in complex codebases**（documentation）：! Wordmark Logo of HumanLayer ./docs/images/wordmark-light.svg 证据：`README.md`
- **Claude Code Go SDK Experimental**（documentation）：A Go SDK for programmatically interacting with Claude Code Anthropic's AI coding assistant . 证据：`claudecode-go/README.md`
- **Go style guidelines**（documentation）：This is the humanlayer Daemon HLD that powers the WUI humanlayer-wui 证据：`hld/CLAUDE.md`
- **HumanLayer Daemon hld**（documentation）：The HumanLayer Daemon hld provides a REST API and JSON-RPC interface for managing Claude Code sessions, approvals, and real-time event streaming. 证据：`hld/README.md`
- **HumanLayer CLI**（documentation）：HumanLayer, but on your command-line. 证据：`hlyr/README.md`
- **Tips and Tricks**（documentation）：This is the HumanLayer Web UI WUI - a desktop application for managing AI agent approvals and sessions. 证据：`humanlayer-wui/CLAUDE.md`
- **humanlayer-wui**（documentation）：Web/desktop UI for the HumanLayer daemon hld built with Tauri and React. 证据：`humanlayer-wui/README.md`
- **APIs**（documentation）：Default to using Bun instead of Node.js. 证据：`scripts/CLAUDE.md`
- **scripts**（documentation）：This project was created using bun init in bun v1.2.23. Bun https://bun.com is a fast all-in-one JavaScript runtime. 证据：`scripts/README.md`
- **HOW TO ITERATE ON GITHUB ACTIONS**（documentation）：1. Select a branch to work on 2. Add branch to workflow trigger : 3. Make one targeted change e.g., cache one thing like Go dependencies 4. Commit and push to the branch ON UPSTREAM git push -u upstream your-debug-branch 5. Check logs with gh : 6. Repeat until working as expected 证据：`.github/workflows/CLAUDE.md`
- **APIs**（documentation）：Default to using Bun instead of Node.js. 证据：`apps/daemon/CLAUDE.md`
- **daemon**（documentation）：This project was created using bun init in bun v1.2.23. Bun https://bun.com is a fast all-in-one JavaScript runtime. 证据：`apps/daemon/README.md`
- **APIs**（documentation）：Default to using Bun instead of Node.js. 证据：`apps/react/CLAUDE.md`
- **bun-react-tailwind-shadcn-template**（documentation）：This project was created using bun init in bun v1.2.23. Bun https://bun.com is a fast all-in-one JavaScript runtime. 证据：`apps/react/README.md`
- **Linear CLI**（documentation）：A command-line interface for interacting with Linear issue tracking. 证据：`hack/linear/README.md`
- **HumanLayer Daemon TypeScript SDK**（documentation）：This SDK provides a TypeScript/JavaScript client for the HumanLayer Daemon HLD REST API. 证据：`hld/sdk/typescript/README.md`
- **WUI Demo Store System**（documentation）：A comprehensive demo store system for creating synthetic product shots and marketing animations using Zustand's slice pattern architecture. 证据：`humanlayer-wui/src/stores/demo/README.md`
- **APIs**（documentation）：Default to using Bun instead of Node.js. 证据：`packages/contracts/CLAUDE.md`
- **@codelayer/protocol**（documentation）：This project was created using bun init in bun v1.2.23. Bun https://bun.com is a fast all-in-one JavaScript runtime. 证据：`packages/contracts/README.md`
- **APIs**（documentation）：Default to using Bun instead of Node.js. 证据：`packages/database/CLAUDE.md`
- **database**（documentation）：This project was created using bun init in bun v1.2.23. Bun https://bun.com is a fast all-in-one JavaScript runtime. 证据：`packages/database/README.md`
- **Package**（package_manifest）：{ "name": "humanlayer", "version": "0.17.2-npm", "description": "HumanLayer, but on your command-line.", "type": "module", "bin": { "humanlayer": "dist/index.js", "hlyr": "dist/index.js", "codelayer": "dist/index.js", "codelayer-nightly": "dist/index.js" }, "files": "dist/ / ", ".claude/ / ", "README.md" , "scripts": { "build": "tsup", "prepack": "npm run build && rimraf .claude && mkdir -p .claude && cp -r ../.claude/commands ../.claude/agents ../.claude/settings.json .claude/", "postpack": "rm -rf .claude", "build:watch": "tsup --watch", "dev": "npm run build && ./dist/index.js", "lint": "eslint . --ext .ts", "format": "prettier --write .", "format:check": "prettier --check .", "test": "v… 证据：`hlyr/package.json`
- **Package**（package_manifest）：{ "name": "humanlayer-wui", "private": true, "version": "0.1.0", "type": "module", "scripts": { "dev": "vite", "build": "tsc && vite build", "preview": "vite preview", "tauri": "tauri", "lint": "eslint . --ext .ts,.tsx", "format": "prettier --write .", "format:check": "prettier --check .", "typecheck": "tsc --noEmit", "check": "bun run format:check && bun run lint && bun run typecheck", "test": "bun test", "storybook": "storybook dev -p 6006", "build-storybook": "storybook build" }, "dependencies": { "@humanlayer/hld-sdk": "file:../hld/sdk/typescript", "@radix-ui/react-checkbox": "^1.3.2", "@radix-ui/react-collapsible": "^1.1.11", "@radix-ui/react-dialog": "^1.1.14", "@radix-ui/react-dropdo… 证据：`humanlayer-wui/package.json`
- **Package**（package_manifest）：{ "name": "codelayer", "private": true, "scripts": { "build": "turbo run build", "dev": "turbo run dev", "lint": "turbo run lint", "format": "prettier --write \" / .{ts,tsx,md}\"", "check-types": "turbo run check-types", "db:generate": "", "db:migrate": "", "db:seed": "" }, "devDependencies": { "@biomejs/biome": "^2.2.6", "prettier": "^3.6.2", "turbo": "^2.5.8", "typescript": "5.9.2" }, "engines": { "node": " =20" }, "packageManager": "bun@1.2.23", "workspaces": "apps/ ", "packages/ " } 证据：`package.json`
- **Package**（package_manifest）：{ "name": "scripts", "private": true, "devDependencies": { "@types/bun": "latest", "@codelayer/contracts": "workspace: " }, "peerDependencies": { "typescript": "^5" } } 证据：`scripts/package.json`
- **Contributing to HumanLayer**（documentation）：If you're looking to contribute, please: 证据：`CONTRIBUTING.md`
- **Package**（package_manifest）：{ "name": "daemon", "module": "index.ts", "type": "module", "private": true, "devDependencies": { "@types/bun": "latest", "bun-types": "^1.2.23" }, "peerDependencies": { "typescript": "^5" }, "dependencies": { "@codelayer/contracts": "workspace: ", "@orpc/openapi": "^1.9.3", "@orpc/server": "^1.9.3", "@orpc/zod": "^1.9.3", "zod": "^4.1.11" } } 证据：`apps/daemon/package.json`
- **Package**（package_manifest）：{ "name": "@codelayer/react", "version": "0.1.0", "private": true, "type": "module", "scripts": { "dev": "PORT=4000 bun --hot src/index.tsx", "start": "NODE ENV=production bun src/index.tsx", "build": "bun run build.ts" }, "dependencies": { "@codelayer/database": "workspace: ", "@electric-sql/react": "^1.0.14", "@radix-ui/react-label": "^2.1.7", "@radix-ui/react-select": "^2.2.5", "@radix-ui/react-slot": "^1.2.3", "@tiptap/extension-collaboration": "^2.2.4", "@tiptap/extension-collaboration-cursor": "^2.2.4", "@tiptap/pm": "^2.2.4", "@tiptap/react": "^2.2.4", "@tiptap/starter-kit": "^2.2.4", "bun-plugin-tailwind": "^0.0.15", "class-variance-authority": "^0.7.1", "clsx": "^2.1.1", "dotenv":… 证据：`apps/react/package.json`
- **Package**（package_manifest）：{ "name": "linear-cli", "version": "1.0.0", "description": "Command line interface for Linear", "main": "./dist/linear-cli.js", "bin": { "linear": "./dist/linear-cli.js" }, "scripts": { "build": "tsc", "start": "ts-node linear-cli.ts", "prepare": "npm run build" }, "dependencies": { "@linear/sdk": "^1.22.0", "chalk": "^4.1.2", "commander": "^9.4.1", "inquirer": "^8.2.5", "node-fetch": "^2.7.0" }, "devDependencies": { "@types/inquirer": "^8.2.5", "@types/node": "^18.11.9", "@types/node-fetch": "^2.6.13", "brace-expansion": " =2.0.2", "ts-node": "^10.9.1", "typescript": "^4.8.4" } } 证据：`hack/linear/package.json`
- **Package**（package_manifest）：{ "name": "@humanlayer/hld-tests", "version": "1.0.0", "private": true, "scripts": { "test:e2e": "bun test-rest-api.ts", "test:e2e:verbose": "VERBOSE=true bun test-rest-api.ts --verbose", "test:e2e:manual": "bun test-rest-api.ts --pause-on-approval" }, "dependencies": { "@types/node": "^20.0.0", "bun": "latest" } } 证据：`hld/e2e/package.json`
- **Package**（package_manifest）：{ "name": "@humanlayer/hld-sdk", "version": "0.1.0", "description": "TypeScript SDK for HumanLayer Daemon REST API", "main": "dist/index.js", "types": "dist/index.d.ts", "scripts": { "build": "bun run tsc", "generate": "./scripts/generate.sh", "test": "bun test", "typecheck": "tsc --noEmit", "check": "bun run typecheck && bun test" }, "dependencies": { "eventsource": "^4.0.0" }, "devDependencies": { "@openapitools/openapi-generator-cli": "^2.7.0", "@types/eventsource": "^1.1.15", "@types/node": "^20.0.0", "@types/node-fetch": "^2.6.12", "typescript": "^5.0.0" } } 证据：`hld/sdk/typescript/package.json`
- **Package**（package_manifest）：{ "name": "@codelayer/contracts", "module": "index.ts", "main": "dist/index.js", "types": "dist/index.d.ts", "type": "module", "scripts": { "build": "bun build --target=node ./src/index.ts --outfile=dist/index.js && bun run build:declaration", "build:declaration": "tsc --emitDeclarationOnly --project tsconfig.types.json", "postbuild": "rimraf tsconfig.types.tsbuildinfo" }, "devDependencies": { "@types/bun": "latest", "bun-types": "^1.2.23", "rimraf": "^6.0.1" }, "peerDependencies": { "typescript": "^5" }, "files": "dist/ .js", "dist/ .d.ts" , "dependencies": { "@orpc/contract": "^1.9.2", "zod": "^4.1.11" }, "exports": { ".": "./src/index.ts", "./daemon": "./src/daemon/index.ts" } } 证据：`packages/contracts/package.json`
- **Package**（package_manifest）：{ "name": "@codelayer/database", "module": "index.ts", "type": "module", "private": true, "scripts": { "db:generate": "drizzle-kit generate", "db:push": "drizzle-kit push", "db:migrate": "bun drizzle-kit migrate", "db:introspect": "drizzle-kit introspect", "db:pull": "drizzle-kit pull", "db:studio": "drizzle-kit studio" }, "devDependencies": { "@types/bun": "latest", "drizzle-kit": "^0.31.5" }, "peerDependencies": { "typescript": "^5" }, "dependencies": { "drizzle-orm": "^0.44.6", "pg": "^8.16.3", "uuidv7": "^1.0.2" } } 证据：`packages/database/package.json`
- **License**（source_file）：Copyright c 2024, humanlayer Authors 证据：`LICENSE`
- **Documentation System**（documentation）：- Version tags follow semver vX.Y.Z - Features added in main branch - Examples updated alongside feature development - Changelog maintained for each version - Both Python and TypeScript packages versioned together - Generate release notes using git commands: - Use git diff v0.5.11..v0.6.0 to see file changes between versions - Always query changes from git before updating CHANGELOG.md - Changelog priorities: - Document API changes first, especially new fields and parameters - Internal changes testing, docs, etc are lower priority - Always document new parameters in models.py or models.ts with their exact names - Link to relevant documentation when adding new features - Changelog organizatio… 证据：`docs/docs.knowledge.md`
- **Case Study: Accelerating Technical Debt Resolution with AI Coding Agents**（documentation）：Case Study: Accelerating Technical Debt Resolution with AI Coding Agents 证据：`docs/case-studies/healthcare-case-study.md`
- **API Reference**（documentation）：Fetch and manage approval requests. 证据：`humanlayer-wui/docs/API.md`
- **Architecture Overview**（documentation）：mermaid graph LR subgraph "Frontend TypeScript " RC React Component RH React Hooks DC Daemon Client end 证据：`humanlayer-wui/docs/ARCHITECTURE.md`
- **Developer Guide**（documentation）：This guide helps you build UI components that interact with the HumanLayer daemon. 证据：`humanlayer-wui/docs/DEVELOPER_GUIDE.md`
- **Hotkey Scope System Documentation**（documentation）：The application uses a hierarchical hotkey scope system built on top of react-hotkeys-hook to properly isolate keyboard shortcuts and prevent conflicts between different UI contexts. 证据：`humanlayer-wui/docs/HOTKEYS.md`
- **What problem s was I solving?**（documentation）：What user-facing changes did I ship? 证据：`.github/PULL_REQUEST_TEMPLATE.md`
- **Development Guide**（documentation）：This guide covers development workflows and tools for the HumanLayer repository. 证据：`DEVELOPMENT.md`
- **HumanLayer Daemon HLD Protocol Documentation**（documentation）：HumanLayer Daemon HLD Protocol Documentation 证据：`hld/PROTOCOL.md`
- **HumanLayer Daemon HLD - TODO**（documentation）：Goal : Reduce N+1 problem for TUI message count display Current issue : TUI needs separate GetConversation call per session to count messages Proposed solution : Add bulk endpoint that returns conversation metadata message count, last message, etc. for multiple sessions Alternative : Extend ListSessions to include conversation metadata Performance impact : Would significantly improve TUI session list load times Files : rpc/handlers.go new endpoint , rpc/types.go new types Priority : Medium - would enable TUI message count feature without performance penalty 证据：`hld/TODO.md`
- **Changelog**（documentation）：All notable changes to the HumanLayer CLI hlyr will be documented in this file. 证据：`hlyr/CHANGELOG.md`
- **HumanLayer Thoughts Management System**（documentation）：HumanLayer Thoughts Management System 证据：`hlyr/THOUGHTS.md`
- **macOS Release Workflow Usage Guide**（documentation）：Testing the Workflow Before Merging 证据：`humanlayer-wui/RELEASE.md`
- **Problems**（documentation）：- 'c' on session table doesn't launch the session creator - selecting "create new session" from the cmd+k launch takes me to a blank screen - search view still doesn't work - the max height should be 80% of the screen height, and only display as many items as can fit - selecting a session from the search view should navigate to the session details 证据：`humanlayer-wui/problems.md`
- **HumanLayer SDK Documentation**（documentation）：🚧 Note : This documentation covers the HumanLayer SDK which is being superseded by CodeLayer. For the latest IDE experience, see the main README. 🚧 证据：`humanlayer.md`
- **Manual Test Plan for Slash Commands Fix**（documentation）：Manual Test Plan for Slash Commands Fix 证据：`test-slash-commands.md`
- **CRITICAL: YOUR ONLY JOB IS TO DOCUMENT AND EXPLAIN THE CODEBASE AS IT EXISTS TODAY**（documentation）：You are a specialist at understanding HOW code works. Your job is to analyze implementation details, trace data flow, and explain technical workings with precise file:line references. 证据：`.claude/agents/codebase-analyzer.md`
- **CRITICAL: YOUR ONLY JOB IS TO DOCUMENT AND EXPLAIN THE CODEBASE AS IT EXISTS TODAY**（documentation）：You are a specialist at finding WHERE code lives in a codebase. Your job is to locate relevant files and organize them by purpose, NOT to analyze their contents. 证据：`.claude/agents/codebase-locator.md`
- **CRITICAL: YOUR ONLY JOB IS TO DOCUMENT AND SHOW EXISTING PATTERNS AS THEY ARE**（documentation）：You are a specialist at finding code patterns and examples in the codebase. Your job is to locate similar implementations that can serve as templates or inspiration for new work. 证据：`.claude/agents/codebase-pattern-finder.md`
- **Core Responsibilities**（documentation）：You are a specialist at extracting HIGH-VALUE insights from thoughts documents. Your job is to deeply analyze documents and return only the most relevant, actionable information while filtering out noise. 证据：`.claude/agents/thoughts-analyzer.md`
- **Core Responsibilities**（documentation）：You are a specialist at finding documents in the thoughts/ directory. Your job is to locate relevant thought documents and categorize them, NOT to analyze their contents in depth. 证据：`.claude/agents/thoughts-locator.md`
- **Core Responsibilities**（documentation）：You are an expert web research specialist focused on finding accurate, relevant information from web sources. Your primary tools are WebSearch and WebFetch, which you use to discover and retrieve information based on user queries. 证据：`.claude/agents/web-search-researcher.md`
- **Commit Changes**（documentation）：You are tasked with creating git commits for the changes made during this session. 证据：`.claude/commands/ci_commit.md`
- **Generate PR Description**（documentation）：You are tasked with generating a comprehensive pull request description following the repository's standard template. 证据：`.claude/commands/ci_describe_pr.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`, `CLAUDE.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`, `CLAUDE.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, humanlayer.md, hld/README.md
- **技术栈**：importance `high`
  - source_paths: hld/go.mod, humanlayer-wui/package.json, hlyr/package.json, packages/database/package.json
- **系统架构**：importance `high`
  - source_paths: hld/daemon/daemon.go, hld/rpc/server.go, hld/api/handlers/server.go, humanlayer-wui/docs/ARCHITECTURE.md
- **事件总线系统**：importance `medium`
  - source_paths: hld/bus/events.go, hld/bus/types.go, hld/bus/events_test.go
- **会话管理**：importance `high`
  - source_paths: hld/session/manager.go, hld/session/types.go, hld/session/claudecode_wrapper.go, hld/session/permission_monitor.go, hld/session/summary.go
- **审批工作流**：importance `high`
  - source_paths: hld/approval/manager.go, hld/approval/types.go, hld/approval/manager_test.go, hld/rpc/approval_handlers.go
- **数据库与存储**：importance `high`
  - source_paths: packages/database/database.ts, packages/database/schema/index.ts, packages/database/schema/thoughts.ts, packages/database/schema/scores.ts, hld/store/sqlite.go
- **REST API**：importance `high`
  - source_paths: hld/api/openapi.yaml, hld/api/handlers/agents.go, hld/api/handlers/sessions.go, hld/api/handlers/approvals.go, hld/api/handlers/files.go

## Repo Inspection Evidence / 源码检查证据

- repo_clone_verified: true
- repo_inspection_verified: true
- repo_commit: `bdea199cec94a1605d2a0de42309d67a14dafdf2`
- inspected_files: `package.json`, `README.md`, `docker-compose.yml`, `docs/workshop.mdx`, `docs/development.mdx`, `docs/package-lock.json`, `docs/mint.json`, `docs/docs.knowledge.md`, `docs/quickstart-typescript.mdx`, `docs/introduction.mdx`, `docs/readme.md`, `docs/case-studies/healthcare-case-study.md`, `packages/database/CLAUDE.md`, `packages/database/index.ts`, `packages/database/types.ts`, `packages/database/package.json`, `packages/database/README.md`, `packages/database/tsconfig.json`, `packages/database/database.ts`, `packages/database/drizzle.config.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

- 当前没有项目专属踩坑日志。宿主 AI 遇到安装、运行、权限、能力边界问题时，必须回到证据来源，不得自行补全。