Keenable Web Search

该技能通过内置的纯标准库 Python 脚本调用无密钥公开搜索端点，按自然语言查询检索网页，支持站点限制与发布时间、索引时间范围过滤，返回带标题、URL 和摘要的排序结果，并可选输出 JSON。适用于研究、新闻、事实核查、文档查找及公司与市场调研；本身不抓取正文，也不处理本地文件。

keenable

热度 16.02

最近更新 2026-06-28 11:55:57

分类：搜索依赖：代码执行 / 第三方 API应用场景：研究 / 市场分析

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add keenable-web-search

Metadata

name: keenable-web-search
version: 1.1.2
description: Web search and research skill. Search the web for current, citable information — news, research papers, documentation, company and market research, fact-checking, and any page beyond the model's training data. Runs a bundled Python script (stdlib only, no API key, no install). NOT for local file operations or non-web tasks.
license: MIT
repository: https://github.com/keenableai/keenable-mcp
homepage: https://keenable.ai
triggers: ["keenable","web search","search the web","search online","look it up","find sources","research","news"]
metadata: {"openclaw":{"emoji":"🔎","tags":["search","web-search","research","news","keyless"]}}

Keenable Web Search 🔎

A web search and research skill. It searches the live web for current, citable information and returns ranked results with titles, URLs, and snippets. A small bundled Python script (standard library only) calls Keenable's keyless public endpoint, so there is no API key and nothing to install.

What this skill is for

Use it whenever the answer depends on information from the web rather than the model's training data:

Research — find papers, primary sources, and background on a topic, then read the most relevant pages before drawing conclusions.
News and current events — look up recent announcements, releases, and developments, with optional date bounds for time-sensitive topics.
Fact-checking and citation — confirm a claim against a primary source and cite the URL.
Company, product, and market research — gather references about an organization, product, technology, or trend.
Documentation lookup — find official docs, changelogs, and references for a library, standard, or API.

Usage

Run the bundled script with a natural-language query (describe the ideal page, not bare keywords):

python3 scripts/search.py "official changelog for the EU AI Act 2025 amendments"

Snippets are starting points. Open the most relevant URL (e.g. with the agent's web-fetch tool) before relying on a claim, and cite the source.

Options

Flag	What it does
`--site <domain>`	Restrict the search to one site, e.g. `--site arxiv.org`.
`--published-after` / `--published-before`	`YYYY-MM-DD` bounds on publish date, for time-sensitive research.
`--acquired-after` / `--acquired-before`	`YYYY-MM-DD` bounds on when the page was indexed.
`--json`	Raw JSON instead of formatted text.

Examples

# News and current events
python3 scripts/search.py "latest OpenAI model announcements"

# Research papers from one site, recent only
python3 scripts/search.py "diffusion models for inverse problems" \
  --site arxiv.org --published-after 2025-01-01

# Documentation lookup
python3 scripts/search.py "Python asyncio TaskGroup reference"

# Machine-readable output for downstream processing
python3 scripts/search.py "EU AI Act timeline" --json

Notes

Keyless. The script calls the public search endpoint and tags traffic with an X-Keenable-Title header; no signup or key is required. The endpoint is rate-limited — on a 429 the script tells you to retry shortly.
No dependencies. Pure Python standard library (urllib), so it runs anywhere Python 3 is available — nothing to pip install.
More about Keenable: keenable.ai.