← Back to Skills Marketplace
256
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install kn70-search
Description
Search the web using multiple engines (Tavily, multi-search-engine, or SearXNG)
README (SKILL.md)
Provides tools for web search using Tavily (AI-optimized), multi-search-engine (17 engines without API key), or SearXNG (self-hosted).
Usage:
// Use Tavily (recommended)
search_web({ query: "AI news", engine: "tavily", max_results: 10 })
// Use multi-search-engine (no API key needed)
search_web({ query: "AI news", engine: "multi", max_results: 10 })
// Use SearXNG (Windows Docker Desktop required)
search_web({ query: "AI news", engine: "searxng", max_results: 10 })
Engine Comparison:
| Engine | Speed | Quality | Cost | Notes |
|---|---|---|---|---|
| tavily | ⚡ Fast | ⭐⭐⭐⭐⭐ | Free (dev) | AI-optimized, includes snippets |
| multi | ⚡ Fast | ⭐⭐⭐⭐ | Free | 17 engines (Baidu, Bing, Google, etc.) |
| searxng | 🐌 Medium | ⭐⭐⭐⭐⭐ | Free | Self-hosted, requires Docker Desktop |
Usage Guidance
Do not install yet. Ask the author to: 1) remove hardcoded API keys and explain any default key (rotate/disable it if it was real); 2) declare required environment variables (TAVILY_API_KEY, SEARXNG_URL) in the manifest; 3) avoid or document absolute exec paths — either include the Python script in the package or replace subprocess calls with in-repo implementations or remote API calls; 4) confirm why api.exec is needed and whether it can be avoided (subprocess execution increases risk); 5) provide a clear install spec or make the skill self-contained and cross-platform. If you must try it, run in a sandboxed environment, do not expose sensitive credentials, and review or remove the embedded key first.
Capability Analysis
Type: OpenClaw Skill
Name: kn70-search
Version: 1.0.0
The skill bundle contains hardcoded absolute Windows file paths (D:\winopenclaw\workspace\...) and attempts to execute external Python scripts via api.exec that are not included in the package (plugin.ts). It also includes a hardcoded Tavily API key (tvly-dev-...) in multiple files (plugin.ts, providers/tavily.ts). The use of api.exec to run unverified local scripts combined with hardcoded credentials and specific environment assumptions is highly irregular and poses a security risk.
Capability Assessment
Purpose & Capability
The skill claims multi-engine web search, which is consistent with the code's intent, but the implementation requires environment variables and external artifacts that are not declared (TAVILY_API_KEY, SEARXNG_URL) and it execs a hardcoded, absolute Windows path to a legacy Python script — behavior that does not align with the manifest (no required env or install).
Instruction Scope
SKILL.md describes engine choices and SearXNG requirements but does not disclose that the runtime will: call external hosts, use an embedded Tavily API key fallback, or execute a Python script at a fixed D:\ path via api.exec. The instructions also omit that api.web_fetch and api.exec will be used, and they don't declare required env vars.
Install Mechanism
There is no install spec, but the code depends on an external Python script at an absolute Windows path (D:\winopenclaw\...), which is not part of the bundle and will fail or cause unexpected behavior. No downloads are declared, but the skill spawns a subprocess — this mismatch is risky and fragile.
Credentials
skill.yaml references a TAVILY_API_KEY env but the registry metadata lists no required envs; the code reads process.env.TAVILY_API_KEY and process.env.SEARXNG_URL and also ships a hardcoded default Tavily API key in multiple files. Requesting/embedding an API key without declaring it and shipping a hardcoded key is disproportionate and may leak credentials.
Persistence & Privilege
The skill does not request always:true and does not modify other skills or system settings. However, it performs subprocess execution (api.exec) and outbound network requests; combined with the other concerns (undeclared envs, hardcoded key, absolute-path exec) this increases potential impact if invoked autonomously.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install kn70-search - After installation, invoke the skill by name or use
/kn70-search - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release of kn70-search web search skill.
- Enables web searches using Tavily (AI-optimized), multi-search-engine (17 engines), or SearXNG (self-hosted).
- Supports multiple engines with customizable results and usage examples.
- Provides an engine comparison table to guide selection.
- No API key required for multi-search-engine; SearXNG setup requires Docker Desktop.
Metadata
Frequently Asked Questions
What is Search?
Search the web using multiple engines (Tavily, multi-search-engine, or SearXNG). It is an AI Agent Skill for Claude Code / OpenClaw, with 256 downloads so far.
How do I install Search?
Run "/install kn70-search" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Search free?
Yes, Search is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Search support?
Search is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Search?
It is built and maintained by hollisken (@hollisken); the current version is v1.0.0.
More Skills