# hermes-agent - 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 hermes-agent 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

- **希望把专业流程带进宿主 AI 的用户**：仓库包含 Skill 文档。 证据：`optional-skills/autonomous-ai-agents/antigravity-cli/SKILL.md`, `optional-skills/autonomous-ai-agents/blackbox/SKILL.md`, `optional-skills/autonomous-ai-agents/grok/SKILL.md`, `optional-skills/autonomous-ai-agents/honcho/SKILL.md` 等 Claim：`clm_0003` 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：`optional-skills/autonomous-ai-agents/antigravity-cli/SKILL.md`, `optional-skills/autonomous-ai-agents/blackbox/SKILL.md`, `optional-skills/autonomous-ai-agents/grok/SKILL.md`, `optional-skills/autonomous-ai-agents/honcho/SKILL.md` 等 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：`AGENTS.md`, `README.md`, `website/i18n/zh-Hans/docusaurus-plugin-content-docs/current/getting-started/quickstart.md` Claim：`clm_0002` supported 0.86

## 怎么开始

- `curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash` 证据：`README.md` Claim：`clm_0004` supported 0.86
- `git clone https://github.com/NousResearch/hermes-agent.git` 证据：`README.md` Claim：`clm_0005` supported 0.86
- `curl -LsSf https://astral.sh/uv/install.sh | sh` 证据：`README.md` Claim：`clm_0006` supported 0.86
- `pip install hermes-agent` 证据：`website/i18n/zh-Hans/docusaurus-plugin-content-docs/current/getting-started/quickstart.md` Claim：`clm_0007` unverified 0.25
- `npm install       # first time` 证据：`AGENTS.md` Claim：`clm_0008` 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**：研究结论、引用和实验结果不能在安装前相信。
- **继续会触碰**：研究判断、命令执行、宿主 AI 配置

### 现在可以相信

- **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：`optional-skills/autonomous-ai-agents/antigravity-cli/SKILL.md`, `optional-skills/autonomous-ai-agents/blackbox/SKILL.md`, `optional-skills/autonomous-ai-agents/grok/SKILL.md`, `optional-skills/autonomous-ai-agents/honcho/SKILL.md` 等 Claim：`clm_0003` 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：`optional-skills/autonomous-ai-agents/antigravity-cli/SKILL.md`, `optional-skills/autonomous-ai-agents/blackbox/SKILL.md`, `optional-skills/autonomous-ai-agents/grok/SKILL.md`, `optional-skills/autonomous-ai-agents/honcho/SKILL.md` 等 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：`AGENTS.md`, `README.md`, `website/i18n/zh-Hans/docusaurus-plugin-content-docs/current/getting-started/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_0004` supported 0.86

### 现在还不能相信

- **研究结论、引用和实验结果不能在安装前相信。**（unverified）：研究 Skill 可以组织问题和路径，但不能替代真实资料检索、论文核验和实验复现。
- **是否适合你的具体研究领域不能直接相信。**（unverified）：Skill 覆盖很多研究主题，不代表对你的领域、资料要求和可信度标准足够。
- **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`, `optional-skills/autonomous-ai-agents/antigravity-cli/SKILL.md`, `optional-skills/autonomous-ai-agents/blackbox/SKILL.md`, `optional-skills/autonomous-ai-agents/grok/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。
- **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

- **研究判断**：问题拆解、资料路径、实验路径、结论结构和可信度判断。 原因：研究型 Skill 可能让输出看起来更专业，但不能替代真实证据核验。
- **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：`AGENTS.md`, `README.md`, `website/i18n/zh-Hans/docusaurus-plugin-content-docs/current/getting-started/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`, `optional-skills/autonomous-ai-agents/antigravity-cli/SKILL.md`, `optional-skills/autonomous-ai-agents/blackbox/SKILL.md`, `optional-skills/autonomous-ai-agents/grok/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：`AGENTS.md`, `README.md`, `website/i18n/zh-Hans/docusaurus-plugin-content-docs/current/getting-started/quickstart.md`
- **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：`website/docs/getting-started/quickstart.md`, `website/docs/user-guide/features/fallback-providers.md`, `website/i18n/zh-Hans/docusaurus-plugin-content-docs/current/user-guide/features/fallback-providers.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**：先验证它能否正确界定研究问题和证据边界，不要先相信研究输出。（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。
- **保留资料和结论核验清单**：如果后续发现引用或实验路径不可靠，可以回到证据边界阶段重新校验。
- **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_0009` 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：`AGENTS.md`, `README.md`, `website/i18n/zh-Hans/docusaurus-plugin-content-docs/current/getting-started/quickstart.md` Claim：`clm_0010` 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：`optional-skills/autonomous-ai-agents/antigravity-cli/SKILL.md`, `optional-skills/autonomous-ai-agents/blackbox/SKILL.md`, `optional-skills/autonomous-ai-agents/grok/SKILL.md`, `optional-skills/autonomous-ai-agents/honcho/SKILL.md` 等 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：`AGENTS.md`, `README.md`, `website/i18n/zh-Hans/docusaurus-plugin-content-docs/current/getting-started/quickstart.md` Claim：`clm_0002` supported 0.86

### 上下文规模

- 文件总数：3208
- 重要文件覆盖：40/3208
- 证据索引条目：180
- 角色 / Skill 条目：170

### 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 hermes-agent 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 hermes-agent 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 hermes-agent 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 170 个角色 / Skill / 项目文档条目。

- **antigravity-cli**（skill）：Operate the Antigravity CLI agy : plugins, auth, sandbox. 激活提示：当用户任务与“antigravity-cli”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：`optional-skills/autonomous-ai-agents/antigravity-cli/SKILL.md`
- **blackbox**（skill）：Delegate coding tasks to Blackbox AI CLI agent. Multi-model agent with built-in judge that runs tasks through multiple LLMs and picks the best result. Requires the blackbox CLI and a Blackbox AI API key. 激活提示：当用户任务与“blackbox”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：`optional-skills/autonomous-ai-agents/blackbox/SKILL.md`
- **grok**（skill）：Delegate coding to xAI Grok Build CLI features, PRs . 激活提示：当用户任务与“grok”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：`optional-skills/autonomous-ai-agents/grok/SKILL.md`
- **honcho**（skill）：Configure and use Honcho memory with Hermes -- cross-session user modeling, multi-profile peer isolation, observation config, dialectic reasoning, session summaries, and context budget enforcement. Use when setting up Honcho, troubleshooting memory, managing profiles with Honcho peers, or tuning observation, recall, and dialectic settings. 激活提示：当用户任务与“honcho”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：`optional-skills/autonomous-ai-agents/honcho/SKILL.md`
- **openhands**（skill）：Delegate coding to OpenHands CLI model-agnostic, LiteLLM . 激活提示：当用户任务与“openhands”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：`optional-skills/autonomous-ai-agents/openhands/SKILL.md`
- **evm**（skill）：Read-only EVM client: wallets, tokens, gas across 8 chains. 激活提示：当用户任务与“evm”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：`optional-skills/blockchain/evm/SKILL.md`
- **hyperliquid**（skill）：Hyperliquid market data, account history, trade review. 激活提示：当用户任务与“hyperliquid”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：`optional-skills/blockchain/hyperliquid/SKILL.md`
- **solana**（skill）：Query Solana blockchain data with USD pricing — wallet balances, token portfolios with values, transaction details, NFTs, whale detection, and live network stats. Uses Solana RPC + CoinGecko. No API key required. 激活提示：当用户任务与“solana”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：`optional-skills/blockchain/solana/SKILL.md`
- **one-three-one-rule**（skill）： 激活提示：当用户任务与“one-three-one-rule”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：`optional-skills/communication/one-three-one-rule/SKILL.md`
- **baoyu-article-illustrator**（skill）：Article illustrations: type × style × palette consistency. 激活提示：当用户任务与“baoyu-article-illustrator”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：`optional-skills/creative/baoyu-article-illustrator/SKILL.md`
- **baoyu-comic**（skill）：Knowledge comics 知识漫画 : educational, biography, tutorial. 激活提示：当用户任务与“baoyu-comic”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：`optional-skills/creative/baoyu-comic/SKILL.md`
- **blender-mcp**（skill）：Control Blender directly from Hermes via socket connection to the blender-mcp addon. Create 3D objects, materials, animations, and run arbitrary Blender Python bpy code. Use when user wants to create or modify anything in Blender. 激活提示：当用户任务与“blender-mcp”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：`optional-skills/creative/blender-mcp/SKILL.md`
- **concept-diagrams**（skill）：Generate flat, minimal light/dark-aware SVG diagrams as standalone HTML files, using a unified educational visual language with 9 semantic color ramps, sentence-case typography, and automatic dark mode. Best suited for educational and non-software visuals — physics setups, chemistry mechanisms, math curves, physical objects aircraft, turbines, smartphones, mechanical watches , anatomy, floor plans, cross-sections, n… 激活提示：当用户任务与“concept-diagrams”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：`optional-skills/creative/concept-diagrams/SKILL.md`
- **ideation**（skill）：Generate project ideas via creative constraints. 激活提示：当用户任务与“ideation”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：`optional-skills/creative/creative-ideation/SKILL.md`
- **hyperframes**（skill）：Create HTML-based video compositions, animated title cards, social overlays, captioned talking-head videos, audio-reactive visuals, and shader transitions using HyperFrames. HTML is the source of truth for video. Use when the user wants a rendered MP4/WebM from an HTML composition, wants to animate text/logos/charts over media, needs captions synced to audio, wants TTS narration, or wants to convert a website into a… 激活提示：当用户任务与“hyperframes”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：`optional-skills/creative/hyperframes/SKILL.md`
- **kanban-video-orchestrator**（skill）：Plan, set up, and monitor a multi-agent video production pipeline backed by Hermes Kanban. Use when the user wants to make ANY video — narrative film, product/marketing, music video, explainer, ASCII/terminal art, abstract/generative loop, comic, 3D, real-time/installation — and the work warrants decomposition into specialized profiles writer, designer, animator, renderer, voice, editor, etc. coordinated through a k… 激活提示：当用户任务与“kanban-video-orchestrator”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：`optional-skills/creative/kanban-video-orchestrator/SKILL.md`
- **meme-generation**（skill）：Generate real meme images by picking a template and overlaying text with Pillow. Produces actual .png meme files. 激活提示：当用户任务与“meme-generation”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：`optional-skills/creative/meme-generation/SKILL.md`
- **pixel-art**（skill）：Pixel art w/ era palettes NES, Game Boy, PICO-8 . 激活提示：当用户任务与“pixel-art”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：`optional-skills/creative/pixel-art/SKILL.md`
- **inference-sh-cli**（skill）：Run 150+ AI apps via inference.sh CLI infsh — image generation, video creation, LLMs, search, 3D, social automation. Uses the terminal tool. Triggers: inference.sh, infsh, ai apps, flux, veo, image generation, video generation, seedream, seedance, tavily 激活提示：当用户任务与“inference-sh-cli”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：`optional-skills/devops/cli/SKILL.md`
- **docker-management**（skill）：Manage Docker containers, images, volumes, networks, and Compose stacks — lifecycle ops, debugging, cleanup, and Dockerfile optimization. 激活提示：当用户任务与“docker-management”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：`optional-skills/devops/docker-management/SKILL.md`
- **hermes-s6-container-supervision**（skill）：Modify, debug, or extend the s6-overlay supervision tree inside the Hermes Agent Docker image — adding new services, debugging profile gateways, understanding the Architecture B main-program pattern. 激活提示：当用户任务与“hermes-s6-container-supervision”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：`optional-skills/devops/hermes-s6-container-supervision/SKILL.md`
- **pinggy-tunnel**（skill）：Zero-install localhost tunnels over SSH via Pinggy. 激活提示：当用户任务与“pinggy-tunnel”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：`optional-skills/devops/pinggy-tunnel/SKILL.md`
- **watchers**（skill）：Poll RSS, JSON APIs, and GitHub with watermark dedup. 激活提示：当用户任务与“watchers”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：`optional-skills/devops/watchers/SKILL.md`
- **adversarial-ux-test**（skill）：Roleplay the most difficult, tech-resistant user for your product. Browse the app as that persona, find every UX pain point, then filter complaints through a pragmatism layer to separate real problems from noise. Creates actionable tickets from genuine issues only. 激活提示：当用户任务与“adversarial-ux-test”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：`optional-skills/dogfood/adversarial-ux-test/SKILL.md`
- **agentmail**（skill）：Give the agent its own dedicated email inbox via AgentMail. Send, receive, and manage email autonomously using agent-owned email addresses e.g. hermes-agent@agentmail.to . 激活提示：当用户任务与“agentmail”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：`optional-skills/email/agentmail/SKILL.md`
- **3-statement-model**（skill）：Build fully-integrated 3-statement models IS, BS, CF in Excel with working capital schedules, D&A roll-forwards, debt schedule, and the plugs that make cash and retained earnings tie. Pairs with excel-author. 激活提示：当用户任务与“3-statement-model”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：`optional-skills/finance/3-statement-model/SKILL.md`
- **comps-analysis**（skill）：Build comparable company analysis in Excel — operating metrics, valuation multiples, statistical benchmarking vs peer sets. Pairs with excel-author. Use for public-company valuation, IPO pricing, sector benchmarking, or outlier detection. 激活提示：当用户任务与“comps-analysis”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：`optional-skills/finance/comps-analysis/SKILL.md`
- **dcf-model**（skill）：Build institutional-quality DCF valuation models in Excel — revenue projections, FCF build, WACC, terminal value, Bear/Base/Bull scenarios, 5x5 sensitivity tables. Pairs with excel-author. Use for intrinsic-value equity analysis. 激活提示：当用户任务与“dcf-model”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：`optional-skills/finance/dcf-model/SKILL.md`
- **excel-author**（skill）：Build auditable Excel workbooks headless with openpyxl — blue/black/green cell conventions, formulas over hardcodes, named ranges, balance checks, sensitivity tables. Use for financial models, audit outputs, reconciliations. 激活提示：当用户任务与“excel-author”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：`optional-skills/finance/excel-author/SKILL.md`
- **lbo-model**（skill）：Build leveraged buyout models in Excel — sources & uses, debt schedule, cash sweep, exit multiple, IRR/MOIC sensitivity. Pairs with excel-author. Use for PE screening, sponsor-case valuation, or illustrative LBO in a pitch. 激活提示：当用户任务与“lbo-model”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：`optional-skills/finance/lbo-model/SKILL.md`
- **merger-model**（skill）：Build accretion/dilution merger models in Excel — pro-forma P&L, synergies, financing mix, EPS impact. Pairs with excel-author. Use for M&A pitches, board materials, or deal evaluation. 激活提示：当用户任务与“merger-model”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：`optional-skills/finance/merger-model/SKILL.md`
- **pptx-author**（skill）：Build PowerPoint decks headless with python-pptx. Pairs with excel-author for model-backed decks where every number traces to a workbook cell. Use for pitch decks, IC memos, earnings notes. 激活提示：当用户任务与“pptx-author”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：`optional-skills/finance/pptx-author/SKILL.md`
- **stocks**（skill）：Stock quotes, history, search, compare, crypto via Yahoo. 激活提示：当用户任务与“stocks”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：`optional-skills/finance/stocks/SKILL.md`
- **minecraft-modpack-server**（skill）：Host modded Minecraft servers CurseForge, Modrinth . 激活提示：当用户任务与“minecraft-modpack-server”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：`optional-skills/gaming/minecraft-modpack-server/SKILL.md`
- **pokemon-player**（skill）：Play Pokemon via headless emulator + RAM reads. 激活提示：当用户任务与“pokemon-player”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：`optional-skills/gaming/pokemon-player/SKILL.md`
- **fitness-nutrition**（skill）： 激活提示：当用户任务与“fitness-nutrition”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：`optional-skills/health/fitness-nutrition/SKILL.md`
- **neuroskill-bci**（skill）： 激活提示：当用户任务与“neuroskill-bci”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：`optional-skills/health/neuroskill-bci/SKILL.md`
- **fastmcp**（skill）：Build, test, inspect, install, and deploy MCP servers with FastMCP in Python. Use when creating a new MCP server, wrapping an API or database as MCP tools, exposing resources or prompts, or preparing a FastMCP server for Claude Code, Cursor, or HTTP deployment. 激活提示：当用户任务与“fastmcp”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：`optional-skills/mcp/fastmcp/SKILL.md`
- **mcporter**（skill）：Use the mcporter CLI to list, configure, auth, and call MCP servers/tools directly HTTP or stdio , including ad-hoc servers, config edits, and CLI/type generation. 激活提示：当用户任务与“mcporter”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：`optional-skills/mcp/mcporter/SKILL.md`
- **openclaw-migration**（skill）：Migrate a user's OpenClaw customization footprint into Hermes Agent. Imports Hermes-compatible memories, SOUL.md, command allowlists, user skills, and selected workspace assets from ~/.openclaw, then reports exactly what could not be migrated and why. 激活提示：当用户任务与“openclaw-migration”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：`optional-skills/migration/openclaw-migration/SKILL.md`
- **huggingface-accelerate**（skill）：Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision FP16/BF16/FP8 . Interactive config, single launch command. HuggingFace ecosystem standard. 激活提示：当用户任务与“huggingface-accelerate”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：`optional-skills/mlops/accelerate/SKILL.md`
- **chroma**（skill）：Open-source embedding database for AI applications. Store embeddings and metadata, perform vector and full-text search, filter by metadata. Simple 4-function API. Scales from notebooks to production clusters. Use for semantic search, RAG applications, or document retrieval. Best for local development and open-source projects. 激活提示：当用户任务与“chroma”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：`optional-skills/mlops/chroma/SKILL.md`
- **clip**（skill）：OpenAI's model connecting vision and language. Enables zero-shot image classification, image-text matching, and cross-modal retrieval. Trained on 400M image-text pairs. Use for image search, content moderation, or vision-language tasks without fine-tuning. Best for general-purpose image understanding. 激活提示：当用户任务与“clip”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：`optional-skills/mlops/clip/SKILL.md`
- **faiss**（skill）：Facebook's library for efficient similarity search and clustering of dense vectors. Supports billions of vectors, GPU acceleration, and various index types Flat, IVF, HNSW . Use for fast k-NN search, large-scale vector retrieval, or when you need pure similarity search without metadata. Best for high-performance applications. 激活提示：当用户任务与“faiss”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：`optional-skills/mlops/faiss/SKILL.md`
- **optimizing-attention-flash**（skill）：Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Use when training/running transformers with long sequences 512 tokens , encountering GPU memory issues with attention, or need faster inference. Supports PyTorch native SDPA, flash-attn library, H100 FP8, and sliding window attention. 激活提示：当用户任务与“optimizing-attention-flash”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：`optional-skills/mlops/flash-attention/SKILL.md`
- **guidance**（skill）：Control LLM output with regex and grammars, guarantee valid JSON/XML/code generation, enforce structured formats, and build multi-step workflows with Guidance - Microsoft Research's constrained generation framework 激活提示：当用户任务与“guidance”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：`optional-skills/mlops/guidance/SKILL.md`
- **huggingface-tokenizers**（skill）：Fast tokenizers optimized for research and production. Rust-based implementation tokenizes 1GB in <20 seconds. Supports BPE, WordPiece, and Unigram algorithms. Train custom vocabularies, track alignments, handle padding/truncation. Integrates seamlessly with transformers. Use when you need high-performance tokenization or custom tokenizer training. 激活提示：当用户任务与“huggingface-tokenizers”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：`optional-skills/mlops/huggingface-tokenizers/SKILL.md`
- **outlines**（skill）：Outlines: structured JSON/regex/Pydantic LLM generation. 激活提示：当用户任务与“outlines”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：`optional-skills/mlops/inference/outlines/SKILL.md`
- **instructor**（skill）：Extract structured data from LLM responses with Pydantic validation, retry failed extractions automatically, parse complex JSON with type safety, and stream partial results with Instructor - battle-tested structured output library 激活提示：当用户任务与“instructor”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：`optional-skills/mlops/instructor/SKILL.md`
- **lambda-labs-gpu-cloud**（skill）：Reserved and on-demand GPU cloud instances for ML training and inference. Use when you need dedicated GPU instances with simple SSH access, persistent filesystems, or high-performance multi-node clusters for large-scale training. 激活提示：当用户任务与“lambda-labs-gpu-cloud”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：`optional-skills/mlops/lambda-labs/SKILL.md`
- **llava**（skill）：Large Language and Vision Assistant. Enables visual instruction tuning and image-based conversations. Combines CLIP vision encoder with Vicuna/LLaMA language models. Supports multi-turn image chat, visual question answering, and instruction following. Use for vision-language chatbots or image understanding tasks. Best for conversational image analysis. 激活提示：当用户任务与“llava”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：`optional-skills/mlops/llava/SKILL.md`
- **modal-serverless-gpu**（skill）：Serverless GPU cloud platform for running ML workloads. Use when you need on-demand GPU access without infrastructure management, deploying ML models as APIs, or running batch jobs with automatic scaling. 激活提示：当用户任务与“modal-serverless-gpu”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：`optional-skills/mlops/modal/SKILL.md`
- **nemo-curator**（skill）：GPU-accelerated data curation for LLM training. Supports text/image/video/audio. Features fuzzy deduplication 16× faster , quality filtering 30+ heuristics , semantic deduplication, PII redaction, NSFW detection. Scales across GPUs with RAPIDS. Use for preparing high-quality training datasets, cleaning web data, or deduplicating large corpora. 激活提示：当用户任务与“nemo-curator”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：`optional-skills/mlops/nemo-curator/SKILL.md`
- **peft-fine-tuning**（skill）：Parameter-efficient fine-tuning for LLMs using LoRA, QLoRA, and 25+ methods. Use when fine-tuning large models 7B-70B with limited GPU memory, when you need to train <1% of parameters with minimal accuracy loss, or for multi-adapter serving. HuggingFace's official library integrated with transformers ecosystem. 激活提示：当用户任务与“peft-fine-tuning”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：`optional-skills/mlops/peft/SKILL.md`
- **pinecone**（skill）：Managed vector database for production AI applications. Fully managed, auto-scaling, with hybrid search dense + sparse , metadata filtering, and namespaces. Low latency <100ms p95 . Use for production RAG, recommendation systems, or semantic search at scale. Best for serverless, managed infrastructure. 激活提示：当用户任务与“pinecone”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：`optional-skills/mlops/pinecone/SKILL.md`
- **pytorch-fsdp**（skill）：Expert guidance for Fully Sharded Data Parallel training with PyTorch FSDP - parameter sharding, mixed precision, CPU offloading, FSDP2 激活提示：当用户任务与“pytorch-fsdp”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：`optional-skills/mlops/pytorch-fsdp/SKILL.md`
- **pytorch-lightning**（skill）：High-level PyTorch framework with Trainer class, automatic distributed training DDP/FSDP/DeepSpeed , callbacks system, and minimal boilerplate. Scales from laptop to supercomputer with same code. Use when you want clean training loops with built-in best practices. 激活提示：当用户任务与“pytorch-lightning”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：`optional-skills/mlops/pytorch-lightning/SKILL.md`
- **qdrant-vector-search**（skill）：High-performance vector similarity search engine for RAG and semantic search. Use when building production RAG systems requiring fast nearest neighbor search, hybrid search with filtering, or scalable vector storage with Rust-powered performance. 激活提示：当用户任务与“qdrant-vector-search”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：`optional-skills/mlops/qdrant/SKILL.md`
- **dspy**（skill）：DSPy: declarative LM programs, auto-optimize prompts, RAG. 激活提示：当用户任务与“dspy”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：`optional-skills/mlops/research/dspy/SKILL.md`
- **sparse-autoencoder-training**（skill）：Provides guidance for training and analyzing Sparse Autoencoders SAEs using SAELens to decompose neural network activations into interpretable features. Use when discovering interpretable features, analyzing superposition, or studying monosemantic representations in language models. 激活提示：当用户任务与“sparse-autoencoder-training”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：`optional-skills/mlops/saelens/SKILL.md`
- **simpo-training**（skill）：Simple Preference Optimization for LLM alignment. Reference-free alternative to DPO with better performance +6.4 points on AlpacaEval 2.0 . No reference model needed, more efficient than DPO. Use for preference alignment when want simpler, faster training than DPO/PPO. 激活提示：当用户任务与“simpo-training”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：`optional-skills/mlops/simpo/SKILL.md`
- **slime-rl-training**（skill）：Provides guidance for LLM post-training with RL using slime, a Megatron+SGLang framework. Use when training GLM models, implementing custom data generation workflows, or needing tight Megatron-LM integration for RL scaling. 激活提示：当用户任务与“slime-rl-training”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：`optional-skills/mlops/slime/SKILL.md`
- **stable-diffusion-image-generation**（skill）：State-of-the-art text-to-image generation with Stable Diffusion models via HuggingFace Diffusers. Use when generating images from text prompts, performing image-to-image translation, inpainting, or building custom diffusion pipelines. 激活提示：当用户任务与“stable-diffusion-image-generation”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：`optional-skills/mlops/stable-diffusion/SKILL.md`
- **tensorrt-llm**（skill）：Optimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency. Use for production deployment on NVIDIA GPUs A100/H100 , when you need 10-100x faster inference than PyTorch, or for serving models with quantization FP8/INT4 , in-flight batching, and multi-GPU scaling. 激活提示：当用户任务与“tensorrt-llm”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：`optional-skills/mlops/tensorrt-llm/SKILL.md`
- **distributed-llm-pretraining-torchtitan**（skill）：Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism FSDP2, TP, PP, CP . Use when pretraining Llama 3.1, DeepSeek V3, or custom models at scale from 8 to 512+ GPUs with Float8, torch.compile, and distributed checkpointing. 激活提示：当用户任务与“distributed-llm-pretraining-torchtitan”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：`optional-skills/mlops/torchtitan/SKILL.md`
- **axolotl**（skill）：Axolotl: YAML LLM fine-tuning LoRA, DPO, GRPO . 激活提示：当用户任务与“axolotl”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：`optional-skills/mlops/training/axolotl/SKILL.md`
- **fine-tuning-with-trl**（skill）：TRL: SFT, DPO, PPO, GRPO, reward modeling for LLM RLHF. 激活提示：当用户任务与“fine-tuning-with-trl”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：`optional-skills/mlops/training/trl-fine-tuning/SKILL.md`
- **unsloth**（skill）：Unsloth: 2-5x faster LoRA/QLoRA fine-tuning, less VRAM. 激活提示：当用户任务与“unsloth”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：`optional-skills/mlops/training/unsloth/SKILL.md`
- **whisper**（skill）：OpenAI's general-purpose speech recognition model. Supports 99 languages, transcription, translation to English, and language identification. Six model sizes from tiny 39M params to large 1550M params . Use for speech-to-text, podcast transcription, or multilingual audio processing. Best for robust, multilingual ASR. 激活提示：当用户任务与“whisper”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：`optional-skills/mlops/whisper/SKILL.md`
- **canvas**（skill）：Canvas LMS integration — fetch enrolled courses and assignments using API token authentication. 激活提示：当用户任务与“canvas”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：`optional-skills/productivity/canvas/SKILL.md`
- **here.now**（skill）：Publish static sites to {slug}.here.now and store private files in cloud Drives for agent-to-agent handoff. 激活提示：当用户任务与“here.now”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：`optional-skills/productivity/here-now/SKILL.md`
- **memento-flashcards**（skill）：- 激活提示：当用户任务与“memento-flashcards”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：`optional-skills/productivity/memento-flashcards/SKILL.md`
- **shop-app**（skill）：Shop.app: product search, order tracking, returns, reorder. 激活提示：当用户任务与“shop-app”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：`optional-skills/productivity/shop-app/SKILL.md`
- **shopify**（skill）：Shopify Admin & Storefront GraphQL APIs via curl. Products, orders, customers, inventory, metafields. 激活提示：当用户任务与“shopify”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：`optional-skills/productivity/shopify/SKILL.md`
- **siyuan**（skill）：SiYuan Note API for searching, reading, creating, and managing blocks and documents in a self-hosted knowledge base via curl. 激活提示：当用户任务与“siyuan”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：`optional-skills/productivity/siyuan/SKILL.md`
- **telephony**（skill）：Give Hermes phone capabilities without core tool changes. Provision and persist a Twilio number, send and receive SMS/MMS, make direct calls, and place AI-driven outbound calls through Bland.ai or Vapi. 激活提示：当用户任务与“telephony”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：`optional-skills/productivity/telephony/SKILL.md`
- **bioinformatics**（skill）：Gateway to 400+ bioinformatics skills from bioSkills and ClawBio. Covers genomics, transcriptomics, single-cell, variant calling, pharmacogenomics, metagenomics, structural biology, and more. Fetches domain-specific reference material on demand. 激活提示：当用户任务与“bioinformatics”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：`optional-skills/research/bioinformatics/SKILL.md`
- **darwinian-evolver**（skill）：Evolve prompts/regex/SQL/code with Imbue's evolution loop. 激活提示：当用户任务与“darwinian-evolver”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：`optional-skills/research/darwinian-evolver/SKILL.md`
- **domain-intel**（skill）：Passive domain reconnaissance using Python stdlib. Subdomain discovery, SSL certificate inspection, WHOIS lookups, DNS records, domain availability checks, and bulk multi-domain analysis. No API keys required. 激活提示：当用户任务与“domain-intel”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：`optional-skills/research/domain-intel/SKILL.md`
- **drug-discovery**（skill）： 激活提示：当用户任务与“drug-discovery”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：`optional-skills/research/drug-discovery/SKILL.md`
- 其余 90 个条目见 `AI_CONTEXT_PACK.json`。

## 证据索引

- 共索引 180 条证据。

- **Hermes Middleware**（documentation）：Hermes middleware is the behavior-changing companion to observer hooks. Observer hooks report what happened. Middleware can change what happens by rewriting a request before execution or by wrapping the execution callback itself. 证据：`docs/middleware/README.md`
- **Hermes Observer Hooks**（documentation）：Hermes observer hooks are the read-only telemetry contract for plugins that need to reconstruct agent execution without changing runtime behavior. This contract supports trace, metrics, audit, replay, and export integrations such as Langfuse, OpenTelemetry-style collectors, and NeMo Relay. 证据：`docs/observability/README.md`
- **Hermes Agent ☤**（documentation）：The self-improving AI agent built by Nous Research https://nousresearch.com . It's the only agent with a built-in learning loop — it creates skills from experience, improves them during use, nudges itself to persist knowledge, searches its own past conversations, and builds a deepening model of who you are across sessions. Run it on a $5 VPS, a GPU cluster, or serverless infrastructure that costs nearly nothing when idle. It's not tied to your laptop — talk to it from Telegram while it works on a cloud VM. 证据：`README.md`
- **providers/**（documentation）：Registry and ABC for every inference provider Hermes knows about. 证据：`providers/README.md`
- **Hermes TUI**（documentation）：React + Ink terminal UI for Hermes. TypeScript owns the screen. Python owns sessions, tools, model calls, and most command logic. 证据：`ui-tui/README.md`
- **Hermes Agent — Web UI**（documentation）：Browser-based dashboard for managing Hermes Agent configuration, API keys, and monitoring active sessions. 证据：`web/README.md`
- **Website**（documentation）：This website is built using Docusaurus https://docusaurus.io/ , a modern static website generator. 证据：`website/README.md`
- **Hermes Desktop ☤**（documentation）：The native desktop app for Hermes Agent ../../README.md — the self-improving AI agent from Nous Research https://nousresearch.com . Same agent, same skills, same memory as the CLI and gateway, in a polished native window — chat with streaming tool output, side-by-side previews, a file browser, voice, and settings, no terminal required. Available for macOS, Windows, and Linux . 证据：`apps/desktop/README.md`
- **SAELens Reference Documentation**（documentation）：This directory contains comprehensive reference materials for SAELens. 证据：`optional-skills/mlops/saelens/references/README.md`
- **{{PROJECT NAME}}**（documentation）：- {{CONCEPT 1}} — {{ONE LINE}} - {{CONCEPT 2}} — {{ONE LINE}} - {{CONCEPT 3}} — {{ONE LINE}} 证据：`optional-skills/software-development/code-wiki/templates/README.md`
- **Readme**（documentation）：Homebrew packaging notes for Hermes Agent. 证据：`packaging/homebrew/README.md`
- **disk-cleanup**（documentation）：Auto-tracks and cleans up ephemeral files created during Hermes Agent sessions — test scripts, temp outputs, cron logs, stale chrome profiles. Scoped strictly to $HERMES HOME and /tmp/hermes- . 证据：`plugins/disk-cleanup/README.md`
- **google meet plugin**（documentation）：Let the hermes agent join a Google Meet call, transcribe it, optionally speak in it, and do the followup work afterwards. 证据：`plugins/google_meet/README.md`
- **Hermes Achievements**（documentation）：Bundled with Hermes Agent. Originally authored by @PCinkusz https://github.com/PCinkusz at https://github.com/PCinkusz/hermes-achievements — vendored into plugins/hermes-achievements/ so it ships with the dashboard out-of-the-box and stays in lockstep with Hermes feature changes. Upstream repo remains the staging ground for new badges and UI iteration. When Hermes is installed via pip install hermes-agent or cloned from source, this plugin auto-registers as a dashboard tab on first hermes dashboard launch. No separate install step. See Built-in Plugins → hermes-achievements ../../website/docs/user-guide/features/built-in-plugins.md in the main docs. 证据：`plugins/hermes-achievements/README.md`
- **ByteRover Memory Provider**（documentation）：Persistent memory via the brv CLI — hierarchical knowledge tree with tiered retrieval fuzzy text → LLM-driven search . 证据：`plugins/memory/byterover/README.md`
- **Hindsight Memory Provider**（documentation）：Long-term memory with knowledge graph, entity resolution, and multi-strategy retrieval. Supports cloud, local embedded, and local external modes. 证据：`plugins/memory/hindsight/README.md`
- **Holographic Memory Provider**（documentation）：Local SQLite fact store with FTS5 search, trust scoring, entity resolution, and HRR-based compositional retrieval. 证据：`plugins/memory/holographic/README.md`
- **Honcho Memory Provider**（documentation）：AI-native cross-session user modeling with multi-pass dialectic reasoning, session summaries, bidirectional peer tools, and persistent conclusions. 证据：`plugins/memory/honcho/README.md`
- **Mem0 Memory Provider**（documentation）：Server-side LLM fact extraction with semantic search, reranking, and automatic deduplication. 证据：`plugins/memory/mem0/README.md`
- **OpenViking Memory Provider**（documentation）：Context database by Volcengine ByteDance with filesystem-style knowledge hierarchy, tiered retrieval, and automatic memory extraction. 证据：`plugins/memory/openviking/README.md`
- **RetainDB Memory Provider**（documentation）：Cloud memory API with hybrid search Vector + BM25 + Reranking and 7 memory types. 证据：`plugins/memory/retaindb/README.md`
- **Supermemory Memory Provider**（documentation）：Semantic long-term memory with profile recall, semantic search, explicit memory tools, and full-session conversation ingest one ingest per session for richer profiles. 证据：`plugins/memory/supermemory/README.md`
- **Model Provider Plugins**（documentation）：Each subdirectory is a self-contained provider profile plugin. The directory layout mirrors plugins/platforms/ : 证据：`plugins/model-providers/README.md`
- **Langfuse Observability Plugin**（documentation）：This plugin ships bundled with Hermes but is opt-in — it only loads when you explicitly enable it. 证据：`plugins/observability/langfuse/README.md`
- **NeMo Relay Observability**（documentation）：Optional Hermes observability plugin that maps Hermes observer hooks to NeMo Relay scopes, LLM spans, tool spans, marks, ATOF, and ATIF. 证据：`plugins/observability/nemo_relay/README.md`
- **security-guidance**（documentation）：Pattern-matched security warnings for code the agent writes. When the agent calls write file , patch , or skill manage with content that matches a known-dangerous code pattern eval, pickle.load, yaml.load, os.system, subprocess with shell=True , dangerouslySetInnerHTML , verify=False , ECB mode, GitHub Actions ${{ github.event. }} injection, torch.load without weights only=True , ... , the plugin appends a warning to the tool's result. The file is still written; the model sees the warning in the next turn and can fix the code or briefly document why the construct is safe. 证据：`plugins/security-guidance/README.md`
- **☤ ASCII Video**（documentation）：Renders any content as colored ASCII character video. Audio, video, images, text, or pure math in, MP4/GIF/PNG sequence out. Full RGB color per character cell, 1080p 24fps default. No GPU. 证据：`skills/creative/ascii-video/README.md`
- **Example Workflows**（documentation）：These are starter API-format workflows for the most common tasks. They're ready to run with scripts/run workflow.py once you've installed or have cloud access to the listed models. 证据：`skills/creative/comfyui/workflows/README.md`
- **Manim Video Skill**（documentation）：Production pipeline for mathematical and technical animations using Manim Community Edition https://www.manim.community/ . 证据：`skills/creative/manim-video/README.md`
- **p5.js Skill**（documentation）：Production pipeline for interactive and generative visual art using p5.js https://p5js.org/ . 证据：`skills/creative/p5js/README.md`
- **LaTeX Templates for ML/AI Conferences**（documentation）：LaTeX Templates for ML/AI Conferences 证据：`skills/research/research-paper-writing/templates/README.md`
- **AAAI 2026 统一LaTeX模板使用说明 / AAAI 2026 Unified LaTeX Template Guide**（documentation）：AAAI 2026 统一LaTeX模板使用说明 / AAAI 2026 Unified LaTeX Template Guide 证据：`skills/research/research-paper-writing/templates/aaai2026/README.md`
- **ACL Paper Styles**（documentation）：This directory contains the latest LaTeX templates for ACL conferences. 证据：`skills/research/research-paper-writing/templates/acl/README.md`
- **Template**（documentation）：Template and style files for CoLM 2025 证据：`skills/research/research-paper-writing/templates/colm2025/README.md`
- **Package**（package_manifest）：{ "name": "hermes-agent", "version": "1.0.0", "description": "An AI agent with advanced tool-calling capabilities, featuring a flexible toolsets system for organizing and managing tools.", "private": true, "workspaces": "apps/ ", "ui-tui", "ui-tui/packages/ ", "web" , "scripts": { "postinstall": "echo '✅ Browser tools ready. Run: python run agent.py --help'", "install:root": "npm install --workspaces=false", "install:web": "npm install --workspace web", "install:tui": "npm install --workspace ui-tui", "install:desktop": "npm install --workspace apps/desktop", "audit:root": "npm audit --workspaces=false", "audit:web": "npm audit --workspace web", "audit:tui": "npm audit --workspace ui-tui",… 证据：`package.json`
- **Package**（package_manifest）：{ "name": "hermes-tui", "version": "0.0.1", "private": true, "type": "module", "scripts": { "dev": "npm run build --prefix packages/hermes-ink && tsx --watch src/entry.tsx", "start": "tsx src/entry.tsx", "build": "node scripts/build.mjs", "type-check": "tsc --noEmit -p tsconfig.json", "lint": "eslint src/ packages/", "lint:fix": "eslint src/ packages/ --fix", "fmt": "prettier --write 'src/ / .{ts,tsx}' 'packages/ / .{ts,tsx}'", "fix": "npm run lint:fix && npm run fmt", "test": "vitest run", "test:watch": "vitest" }, "dependencies": { "@hermes/ink": "file:./packages/hermes-ink", "@nanostores/react": "^1.1.0", "ink": "^6.8.0", "ink-text-input": "^6.0.0", "nanostores": "^1.2.0", "react": "^19.… 证据：`ui-tui/package.json`
- **Package**（package_manifest）：{ "name": "web", "private": true, "version": "0.0.0", "type": "module", "scripts": { "dev": "vite", "build": "tsc -b && vite build", "lint": "eslint .", "preview": "vite preview" }, "dependencies": { "@nous-research/ui": "0.18.2", "@observablehq/plot": "^0.6.17", "@react-three/fiber": "^9.6.0", "@tailwindcss/vite": "^4.2.1", "@xterm/addon-fit": "^0.11.0", "@xterm/addon-unicode11": "^0.9.0", "@xterm/addon-web-links": "^0.12.0", "@xterm/addon-webgl": "^0.19.0", "@xterm/xterm": "^6.0.0", "class-variance-authority": "^0.7.1", "clsx": "^2.1.1", "gsap": "^3.15.0", "leva": "^0.10.1", "lucide-react": "^0.577.0", "motion": "^12.38.0", "qrcode": "^1.5.4", "react": "^19.2.4", "react-dom": "^19.2.4", "… 证据：`web/package.json`
- **Package**（package_manifest）：{ "name": "website", "version": "0.0.0", "private": true, "scripts": { "docusaurus": "docusaurus", "prestart": "node scripts/prebuild.mjs", "start": "docusaurus start", "prebuild": "node scripts/prebuild.mjs", "build": "docusaurus build", "swizzle": "docusaurus swizzle", "deploy": "docusaurus deploy", "clear": "docusaurus clear", "serve": "docusaurus serve", "write-translations": "docusaurus write-translations", "write-heading-ids": "docusaurus write-heading-ids", "typecheck": "tsc", "lint:diagrams": "ascii-guard lint --exclude-code-blocks docs" }, "dependencies": { "@docusaurus/core": "3.9.2", "@docusaurus/preset-classic": "3.9.2", "@docusaurus/theme-mermaid": "^3.9.2", "@easyops-cn/docusa… 证据：`website/package.json`
- **Package**（package_manifest）：{ "name": "@hermes/bootstrap-installer", "private": true, "version": "0.0.1", "description": "Hermes Setup — signed installer that drives scripts/install.ps1 with a polished native UI.", "type": "module", "scripts": { "dev": "vite --host 127.0.0.1 --port 5175", "build": "tsc -b && vite build", "preview": "vite preview", "tauri": "tauri", "tauri:dev": "tauri dev", "tauri:build": "tauri build", "tauri:build:debug": "tauri build --debug" }, "dependencies": { "@nous-research/ui": "0.16.0", "@tailwindcss/vite": "^4.2.1", "@tailwindcss/typography": "^0.5.19", "@tauri-apps/api": "^2.0.0", "@tauri-apps/plugin-dialog": "^2.0.0", "@tauri-apps/plugin-opener": "^2.0.0", "@tauri-apps/plugin-process": "^… 证据：`apps/bootstrap-installer/package.json`
- **Package**（package_manifest）：{ "name": "hermes", "productName": "Hermes", "private": true, "version": "0.15.1", "description": "Native desktop shell for Hermes Agent.", "author": "Nous Research", "type": "module", "main": "electron/main.cjs", "engines": { "node": "^20.19.0 =22.12.0" }, "scripts": { "dev": "concurrently -k \"npm:dev:renderer\" \"npm:dev:electron\"", "dev:fake-boot": "cross-env HERMES DESKTOP BOOT FAKE=1 HERMES DESKTOP BOOT FAKE STEP MS=650 npm run dev", "dev:renderer": "node scripts/assert-root-install.cjs && vite --host 127.0.0.1 --port 5174", "dev:electron": "wait-on http://127.0.0.1:5174 && cross-env XCURSOR SIZE=24 HERMES DESKTOP DEV SERVER=http://127.0.0.1:5174 electron .", "profile:main": "wait-on… 证据：`apps/desktop/package.json`
- **Package**（package_manifest）：{ "name": "@hermes/shared", "private": true, "version": "0.0.0", "type": "module", "exports": { ".": "./src/index.ts" }, "types": "./src/index.ts", "scripts": { "type-check": "tsc -p tsconfig.json --noEmit" }, "devDependencies": { "typescript": "^6.0.3" } } 证据：`apps/shared/package.json`
- **Package**（package_manifest）：{ "name": "hermes-whatsapp-bridge", "version": "1.0.0", "description": "WhatsApp bridge for Hermes Agent using Baileys", "private": true, "type": "module", "scripts": { "start": "node bridge.js" }, "dependencies": { "@whiskeysockets/baileys": "WhiskeySockets/Baileys 01047debd81beb20da7b7779b08edcb06aa03770", "express": "^4.21.0", "qrcode-terminal": "^0.12.0", "pino": "^9.0.0" }, "overrides": { "protobufjs": "^7.5.5" } } 证据：`scripts/whatsapp-bridge/package.json`
- **Package**（package_manifest）：{ "name": "@hermes/ink", "version": "0.0.1", "private": true, "type": "module", "scripts": { "build": "esbuild src/entry-exports.ts --bundle --platform=node --format=esm --packages=external --outdir=dist" }, "sideEffects": true, "main": "./index.js", "types": "./index.d.ts", "exports": { ".": { "types": "./index.d.ts", "import": "./index.js", "default": "./index.js" }, "./text-input": { "types": "./text-input.d.ts", "import": "./text-input.js", "default": "./text-input.js" }, "./package.json": "./package.json" }, "peerDependencies": { "ink-text-input": " =6.0.0", "react": " =19.0.0" }, "dependencies": { "@alcalzone/ansi-tokenize": "^0.1.0", "auto-bind": "^5.0.0", "bidi-js": "^1.0.0", "chalk… 证据：`ui-tui/packages/hermes-ink/package.json`
- **Contributing**（documentation）：Thank you for contributing to Hermes Agent! This guide covers setting up your dev environment, understanding the codebase, and getting your PR merged. 证据：`website/docs/developer-guide/contributing.md`
- **ComfyUI Skill Tests**（documentation）：Pytest suite covering the skill's scripts. Pure-stdlib unit tests run without any setup; cloud integration tests need a Comfy Cloud API key. 证据：`skills/creative/comfyui/tests/README.md`
- **Hermes Agent - Development Guide**（documentation）：Instructions for AI coding assistants and developers working on the hermes-agent codebase. 证据：`AGENTS.md`
- **Design System: Claude Anthropic**（documentation）：Hermes Agent — Implementation Notes The original site uses proprietary fonts. For self-contained HTML output, use these CDN substitutes: - Primary: Inter Mono: JetBrains Mono - Font stack CSS : font-family: 'Inter', system-ui, -apple-system, 'Segoe UI', Roboto, sans-serif; - Mono stack CSS : font-family: 'JetBrains Mono', ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, 'Liberation Mono', 'Courier New', monospace; Use write file to create HTML, serve via generative-widgets skill cloudflared tunnel . Verify visual accuracy with browser vision after generating. 证据：`skills/creative/popular-web-designs/templates/claude.md`
- **Contributing to Hermes Agent**（documentation）：Thank you for contributing to Hermes Agent! This guide covers everything you need: setting up your dev environment, understanding the architecture, deciding what to build, and getting your PR merged. 证据：`CONTRIBUTING.md`
- **Antigravity CLI agy**（skill_instruction）：Operator guide for the Antigravity CLI, invoked as agy . Run all agy commands through the Hermes terminal tool; inspect its config and logs with read file . This skill is reference + procedure — it does not wrap a network API, so there is nothing to authenticate from Hermes itself. 证据：`optional-skills/autonomous-ai-agents/antigravity-cli/SKILL.md`
- **Blackbox CLI**（skill_instruction）：Delegate coding tasks to Blackbox AI https://www.blackbox.ai/ via the Hermes terminal. Blackbox is a multi-model coding agent CLI that dispatches tasks to multiple LLMs Claude, Codex, Gemini, Blackbox Pro and uses a judge to select the best implementation. 证据：`optional-skills/autonomous-ai-agents/blackbox/SKILL.md`
- **Grok Build CLI — Hermes Orchestration Guide**（skill_instruction）：Grok Build CLI — Hermes Orchestration Guide 证据：`optional-skills/autonomous-ai-agents/grok/SKILL.md`
- **Honcho Memory for Hermes**（skill_instruction）：Honcho provides AI-native cross-session user modeling. It learns who the user is across conversations and gives every Hermes profile its own peer identity while sharing a unified view of the user. 证据：`optional-skills/autonomous-ai-agents/honcho/SKILL.md`
- **OpenHands CLI**（skill_instruction）：Delegate coding tasks to the OpenHands CLI https://github.com/All-Hands-AI/OpenHands via the terminal tool. OpenHands is model-agnostic: any LiteLLM-supported provider OpenAI, Anthropic, OpenRouter, DeepSeek, Ollama, vLLM, etc. . 证据：`optional-skills/autonomous-ai-agents/openhands/SKILL.md`
- **EVM Blockchain Skill**（skill_instruction）：Query EVM-compatible blockchain data across 8 chains with USD pricing. 14 commands: wallet portfolio, token info, transactions, activity, gas tracker, network stats, price lookup, multi-chain scan, whale detection, ENS resolution, allowance checker, contract inspector, and transaction decoder. 证据：`optional-skills/blockchain/evm/SKILL.md`
- **Hyperliquid Skill**（skill_instruction）：Query Hyperliquid market and account data through the public /info endpoint. Read-only — no API key, no signing, no order placement. 证据：`optional-skills/blockchain/hyperliquid/SKILL.md`
- **Solana Blockchain Skill**（skill_instruction）：Query Solana on-chain data enriched with USD pricing via CoinGecko. 8 commands: wallet portfolio, token info, transactions, activity, NFTs, whale detection, network stats, and price lookup. 证据：`optional-skills/blockchain/solana/SKILL.md`
- **1-3-1 Communication Rule**（skill_instruction）：Structured decision-making format for when a task has multiple viable approaches and the user needs a clear recommendation. Produces a concise problem framing, three options with trade-offs, and an actionable plan for the recommended path. 证据：`optional-skills/communication/one-three-one-rule/SKILL.md`
- **Article Illustrator**（skill_instruction）：Adapted from baoyu-article-illustrator https://github.com/JimLiu/baoyu-skills for Hermes Agent's tool ecosystem. 证据：`optional-skills/creative/baoyu-article-illustrator/SKILL.md`
- **Knowledge Comic Creator**（skill_instruction）：Adapted from baoyu-comic https://github.com/JimLiu/baoyu-skills for Hermes Agent's tool ecosystem. 证据：`optional-skills/creative/baoyu-comic/SKILL.md`
- **Blender MCP**（skill_instruction）：Control a running Blender instance from Hermes via socket on TCP port 9876. 证据：`optional-skills/creative/blender-mcp/SKILL.md`
- 其余 120 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/middleware/README.md`, `docs/observability/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/middleware/README.md`, `docs/observability/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。

- **Hermes Agent Overview & System Architecture**：importance `high`
  - source_paths: README.md, run_agent.py, cli.py, hermes_bootstrap.py, hermes_cli/main.py
- **Memory, Skills & Self-Improving Learning Loop**：importance `high`
  - source_paths: agent/memory_manager.py, agent/memory_provider.py, agent/curator.py, agent/curator_backup.py, agent/manual_compression_feedback.py
- **Multi-Agent Architecture, Delegation & Protocol Integration**：importance `high`
  - source_paths: tools/delegate_tool.py, tools/mixture_of_agents_tool.py, tools/mcp_tool.py, tools/mcp_oauth.py, tools/mcp_oauth_manager.py
- **Frontend Surfaces, Themes & Platform Integrations**：importance `high`
  - source_paths: web/src/App.tsx, web/src/themes/presets.ts, web/src/themes/context.tsx, web/src/themes/types.ts, web/src/i18n/index.ts

## Repo Inspection Evidence / 源码检查证据

- repo_clone_verified: true
- repo_inspection_verified: true
- repo_commit: `846821d8c0b57808a24ef8f3de21e072c573db1d`
- inspected_files: `uv.lock`, `Dockerfile`, `pyproject.toml`, `README.md`, `package.json`, `docker-compose.yml`, `docs/middleware/README.md`, `docs/security/network-egress-isolation.md`, `docs/observability/README.md`, `docs/kanban/multi-gateway.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: 来源证据：[Bug]: Fails to install NeuTTS during setup

- 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：[Bug]: Fails to install NeuTTS during setup
- Host AI rule: 来源问题仍为 open，Pack Agent 需要复核是否仍影响当前版本。
- Why it matters: 可能阻塞安装或首次运行。
- Evidence: community_evidence:github | cevd_ebdf0d00876046d89748da04ed227c7e | https://github.com/NousResearch/hermes-agent/issues/3002 | 来源讨论提到 python 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 2: 来源证据：[skills-index-watchdog] Skills index is stale or degraded (degraded)

- Trigger: GitHub 社区证据显示该项目存在一个配置相关的待验证问题：[skills-index-watchdog] Skills index is stale or degraded (degraded)
- Host AI rule: 来源问题仍为 open，Pack Agent 需要复核是否仍影响当前版本。
- Why it matters: 可能增加新用户试用和生产接入成本。
- Evidence: community_evidence:github | cevd_0b1cf269e8544779854a23690e9187be | https://github.com/NousResearch/hermes-agent/issues/38240 | 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 3: 来源证据：[Bug]: Desktop "Remote gateway" can't connect over WebSocket — /api/ws always rejected (4403) for packaged Electron cli…

- Trigger: GitHub 社区证据显示该项目存在一个安全/权限相关的待验证问题：[Bug]: Desktop "Remote gateway" can't connect over WebSocket — /api/ws always rejected (4403) for packaged Electron client
- Host AI rule: 来源问题仍为 open，Pack Agent 需要复核是否仍影响当前版本。
- Why it matters: 可能影响授权、密钥配置或安全边界。
- Evidence: community_evidence:github | cevd_533af3ccff3243efbc8a4a049fb486f8 | https://github.com/NousResearch/hermes-agent/issues/38412 | 来源讨论提到 python 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 | github_repo:1024554267 | https://github.com/NousResearch/hermes-agent | host_targets=hermes, claude_code, claude, chatgpt
- 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 | github_repo:1024554267 | https://github.com/NousResearch/hermes-agent | 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: 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 | github_repo:1024554267 | https://github.com/NousResearch/hermes-agent | 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。

### Constraint 7: 下游验证发现风险项

- Trigger: no_demo
- Host AI rule: 进入安全/权限治理复核队列。
- Why it matters: 下游已经要求复核，不能在页面中弱化。
- Evidence: downstream_validation.risk_items | github_repo:1024554267 | https://github.com/NousResearch/hermes-agent | 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 8: 存在评分风险

- Trigger: no_demo
- Host AI rule: 把风险写入边界卡，并确认是否需要人工复核。
- Why it matters: 风险会影响是否适合普通用户安装。
- Evidence: risks.scoring_risks | github_repo:1024554267 | https://github.com/NousResearch/hermes-agent | 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 9: 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 | github_repo:1024554267 | https://github.com/NousResearch/hermes-agent | 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 10: 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 | github_repo:1024554267 | https://github.com/NousResearch/hermes-agent | 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。