據 AlphaMoat 統計,截至 2026-09,Querit Search 累計下載 2,632 次、獲 3 個星標,在全部 5,078 個 AI 技能中排名第 3,996 位,在「知識管理」分類(442 個)中排名第 345 位。
| 平台 | 累計安裝 | 累計下載 | 星標 | 評論 |
|---|---|---|---|---|
| ClawHub ↗ | 81 | 2475 | 3 | - |
| SkillHub ↗ | 142 | 2632 | 3 | — |
各平台的計數口徑不同,統計時間不一致。
搜尋透過 Querit API 進行,需要配置 QUERIT_API_KEY;文件所述查詢長度上限為 72 個字元。網頁內容提取會直接訪問使用者提供的 URL,並非經 Querit 搜尋 API 獲取。訪問非公開 URL 前應核對目的地址和許可權;外部搜尋結果需回到原始來源核實。
資料釋出於 2026-10-29
以下是根據技能文件生成的示例,不代表實際執行結果。
2026-09 起統計方法調整:同名技能按作者分別統計,並覆蓋全部技能。圖中虛線為調整前的月份,與之後的資料不直接比較,也不計算跨越調整的增長。
| 月份 | 累計下載 | 環比 | 星標 |
|---|---|---|---|
| 2026-09 | 2632 | — | 3 |
| 2026-08 | 2400 | +3.2% | 3 |
| 2026-07 | 2326 | +7.1% | 3 |
| 2026-06 | 2172 | +4.7% | 3 |
| 2026-05 | 2075 | — | 3 |
| 2026-03 | 1832 | — | 3 |
以下是作者釋出的原始檔案,按純文字顯示,本站未做修改或稽核;其中的連結不可點選,命令請核實後再執行。抓取於 2026-10-23。
---
name: querit-search
description: >-
Web search via Querit.ai API. Use when you need to search the web for
documentation, current events, facts, or any web content. Returns
structured results with titles, URLs, and snippets.
metadata: {"openclaw":{"emoji":"🔎","requires":{"env":["QUERIT_API_KEY"]},"primaryEnv":"QUERIT_API_KEY","install":[{"id":"node","kind":"node","label":"Install npm dependencies"}]}}
---
# Querit Search
Web search and content extraction via the Querit.ai API. No browser required.
## Setup
Needs env: `QUERIT_API_KEY` — get a free key at https://querit.ai (1,000 queries/month).
## Search
```bash
node {baseDir}/search.js "query" # 5 results (default)
node {baseDir}/search.js "query" -n 10 # more results (max 100)
node {baseDir}/search.js "query" --lang english # language filter
node {baseDir}/search.js "query" --country "united states" # country filter
node {baseDir}/search.js "query" --date w1 # past week (d1/w1/m1/y1)
node {baseDir}/search.js "query" --site-include github.com # only this domain
node {baseDir}/search.js "query" --site-exclude reddit.com # exclude domain
node {baseDir}/search.js "query" --content # also extract page content
node {baseDir}/search.js "query" --json # raw JSON output
```
Flags can be combined:
```bash
node {baseDir}/search.js "react hooks" -n 3 --lang english --site-include reactjs.org --content
```
## Extract Page Content
```bash
node {baseDir}/content.js https://example.com/article
```
Fetches a URL and extracts the main readable content as markdown.
## Output Format
### Search results (default)
```
1. Page Title
https://example.com/page
Site: example.com
Age: 3 days ago
Description snippet from search results
2. Another Page
...
```
### With --content
After the result listing, each page's extracted markdown content is appended:
```
### 1. Page Title
URL: https://example.com/page
# Extracted heading
Extracted body content in markdown...
---
```
### With --json
Raw JSON array of result objects with fields: `url`, `title`, `snippet`, `page_age`, `page_time`.
## When to Use
- Searching for documentation, API references, or tutorials
- Looking up facts, current events, or recent information
- Finding content from specific websites (use `--site-include`)
- Fetching and reading a web page's content (use `--content` or `content.js`)
- Any task requiring web search without interactive browsing
## Limitations
- Query limited to 72 characters (auto-truncated with warning)
- Max 100 results per query
- Max 20 domains per site filter
- Free tier: 1,000 queries/month, 1 QPS
- Supported languages: english, japanese, korean, german, french, spanish, portuguese
把長文字、文件與網頁內容壓縮成要點和關鍵詞的摘要工具。
透過 IMA OpenAPI 管理筆記與知識庫內容。
規定如何為網頁請求選擇合適工具的必讀指引。
呼叫百度 AI 搜尋介面做網頁檢索的技能。
用 uvx 直接呼叫 markitdown,把各類檔案轉成 Markdown。
面向 OpenClaw 的預設網頁搜尋,即時檢索多個引擎。
用 DuckDuckGo 介面做網頁、新聞、圖片與影片檢索的技能。
檢索並發現可用 OpenClaw 技能的技能。
資料月份:2026-09 · 下載/星標/安裝數聚合自公開技能註冊庫(ClawHub、SkillHub),按月更新。口徑說明見資料方法。