OpenClaw Docs Search

Search OpenClaw official docs and return concise Markdown. Invoke when users ask about OpenClaw docs, config, CLI, channels, or skills.

Security Scan

The skill's code, instructions, and dependencies are consistent with its stated purpose (search and fetch OpenClaw docs), with only minor operational notes about where searches are sent and Node engine expectations.

Purpose & Capability

Name and description match the included code: the package searches documentation and fetches pages to extract #content-area and convert to Markdown. One minor mismatch: the SKILL.md says it queries the 'official search API', but the code posts to a third-party search endpoint (https://leaves.mintlify.com/api/search/clawdhub). Using a hosted search provider for docs is plausible but worth noting.

Instruction Scope

SKILL.md instructions are narrowly scoped: install dependencies in the skill workspace and run the provided node CLI to 'search' or 'doc'. The instructions do not direct reading unrelated local files or exfiltrating secrets. They explicitly constrain behavior (no bulk mirroring, avoid sensitive pages).

Install Mechanism

There is no registry install spec; the SKILL.md instructs a local 'npm install' in ~/.openclaw/workspace/skills/openclaw-docs-search. Dependencies come from npm (jsdom, turndown) — a standard approach. Note: several packaged dependencies list Node engine requirements >=20.19 in package-lock, while SKILL.md recommends Node >=18; this could cause install/runtime issues on older Node versions.

Credentials

The skill requests no environment variables, credentials, or config paths. The code does not read local secrets or other configs; it only issues HTTP requests to the search endpoint and to docs.openclaw.ai pages, which is proportionate for a doc search skill.

Persistence & Privilege

No elevated privileges are requested. always is false, and the skill is user-invocable. It does not attempt to modify other skills or system-wide config.

Assessment

This package appears to do what it says: search OpenClaw docs and return Markdown. Before installing, note two practical points: (1) search queries are sent to a third-party host (leaves.mintlify.com) rather than directly to docs.openclaw.ai — if your queries contain sensitive, private, or internal data avoid sending them; (2) some dependencies in package-lock specify Node >=20.19 whereas the SKILL.md suggests Node >=18, so you may need a newer Node runtime to install/run successfully. If you want to proceed: inspect the included dist/search.js (already bundled), run npm install in an isolated environment or sandbox, and optionally review network traffic to confirm requests only go to the expected endpoints (leaves.mintlify.com and docs.openclaw.ai). If you need stronger assurance, ask the publisher to document the search endpoint explicitly or to host a search endpoint under docs.openclaw.ai.

用于检索 OpenClaw 官方文档，并将结果整理为适合 LLM 阅读的 Markdown。

⚡ 快速使用（必读）

前置条件：安装依赖

首次使用前需要安装依赖：

cd ~/.openclaw/workspace/skills/openclaw-docs-search
npm install

命令一：搜索文档

node ~/.openclaw/workspace/skills/openclaw-docs-search/dist/search.js search "<英文关键字>" en

• 英文搜索（推荐）: ... search "multi agent" en
• 中文搜索: ... search "多代理" zh-Hans

示例：

node ~/.openclaw/workspace/skills/openclaw-docs-search/dist/search.js search "create multiple agents" en

返回格式化的 Markdown 搜索结果，包含页面路径（page 字段），后续可用于获取详情。

命令二：获取文档详情

从搜索结果中获取 page 路径后，使用此命令获取完整内容：

node ~/.openclaw/workspace/skills/openclaw-docs-search/dist/search.js doc "<page路径>"

示例：

node ~/.openclaw/workspace/skills/openclaw-docs-search/dist/search.js doc "concepts/multi-agent"

返回干净的 Markdown 正文（自动提取 #content-area 并转换为 Markdown）。

---

适用场景

• 用户询问 OpenClaw 的安装、配置、CLI、渠道接入、Skills、Gateway、诊断或运维问题
• 需要快速定位某个官方文档页面并返回简洁摘要
• 需要进一步读取单篇文档正文，并尽量减少噪音与 Token 消耗

⚠️ 重要提示： 由于官方英文文档的搜索命中率远高于中文文档，请在发起搜索前，务必将用户的中文查询意图翻译为英文关键字进行搜索。接口默认语言已设置为 en。如果确实需要查阅中文结果，再显式将 language 指定为 zh-Hans。

---

工作原理

搜索流程

1. 向 OpenClaw 官方搜索 API 发送请求
2. 获取 JSON 结果，提取 page、breadcrumbs、content 等字段
3. 清理 HTML 标签（如 <mark> 高亮），格式化为 Markdown

返回示例：

### 1. Agents > Multi-agent > Multi-Agent Routing
- **路径**: `concepts/multi-agent`
- **内容**: Multi-Agent Routing Goal: multiple isolated agents...

获取详情流程

1. 拼接 URL：https://docs.openclaw.ai/<page路径>
2. 获取 HTML 页面
3. 提取 id="content-area" 元素
4. 使用 turndown 转换为 Markdown
5. 添加来源标识：> 来源: https://docs.openclaw.ai/...

输出要求

• 搜索结果优先返回精简 Markdown，而不是原始 JSON
• 删除搜索高亮标签、冗余元数据、重复标题和无关噪音
• 详情页只保留 #content-area 主内容，不返回整页导航、页脚或脚本内容
• 在结果中尽量保留来源路径，便于后续继续深挖

约束

• 不批量镜像整个站点
• 不请求与文档正文无关的敏感页面
• 不返回真实密钥、令牌或用户私密信息
• 优先以"按需查阅"的方式工作，减少不必要请求

常见问题

搜索无结果？

1. 尝试用更通用的英文关键词
2. 使用官方文档目录探索：https://docs.openclaw.ai/llms.txt

依赖安装失败？

确保 Node.js 版本 >= 18，然后重试：

cd ~/.openclaw/workspace/skills/openclaw-docs-search
rm -rf node_modules package-lock.json
npm install