w
zhipu web search
Scanned@whyhit2005
npx machina-cli add skill @whyhit2005/zhipu-web-search --openclawFiles (1)
SKILL.md
4.0 KB
Zhipu Search
Web search via Zhipu AI API, supporting multiple search engines and flexible parameter configuration.
Quick Start
Basic Search
# Use default parameters
search_query = "OpenClaw latest version"
search_engine = "search_std"
Advanced Search (Full Parameters)
search_query = "AI development trends" # Required, max 70 chars
search_engine = "search_pro" # Required: search_std/search_pro/search_pro_sogou/search_pro_quark
search_intent = true # Optional, default false, enable search intent recognition
count = 20 # Optional, default 10, range 1-50
search_domain_filter = "example.com" # Optional, whitelist domain filter
search_recency_filter = "oneWeek" # Optional: oneDay/oneWeek/oneMonth/oneYear/noLimit
content_size = "high" # Optional: medium/high, control content detail level
request_id = "unique-request-id" # Optional, unique request identifier
user_id = "user-123456" # Optional, end user ID (6-128 chars)
Usage Methods
Method 1: Direct Script Call (Recommended)
python scripts/zhipu_search.py \
--query "search content" \
--engine search_pro \
--count 10
Method 2: Use OpenClaw Tool
System will automatically select appropriate parameters based on needs.
API Parameter Reference
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
| search_query | string | ✅ | - | Search content, recommended ≤70 chars |
| search_engine | enum | ✅ | - | search_std/search_pro/search_pro_sogou/search_pro_quark |
| search_intent | boolean | - | false | Enable search intent recognition |
| count | integer | - | 10 | Result count, 1-50 |
| search_domain_filter | string | - | - | Whitelist domain filter |
| search_recency_filter | enum | - | noLimit | oneDay/oneWeek/oneMonth/oneYear/noLimit |
| content_size | enum | - | - | medium/high, control content length |
| request_id | string | - | - | Unique request identifier |
| user_id | string | - | - | End user ID (6-128 chars) |
Search Engine Selection Guide
| Engine | Use Case |
|---|---|
| search_std | Basic search, regular Q&A |
| search_pro | Advanced search, need more accurate results |
| search_pro_sogou | Sogou search, China domestic content |
| search_pro_quark | Quark search, specific scenarios |
Response Structure
{
"id": "task-id",
"created": 1704067200,
"request_id": "request-id",
"search_intent": [
{
"query": "original query",
"intent": "SEARCH_ALL",
"keywords": "rewritten keywords"
}
],
"search_result": [
{
"title": "title",
"content": "content summary",
"link": "result link",
"media": "site name",
"icon": "site icon",
"refer": "reference number",
"publish_date": "publish date"
}
]
}
Environment Requirements
- Environment variable
ZHIPU_API_KEYmust be configured - Python 3.7+
- requests library
Notes
- search_query should be kept within 70 characters
- search_pro_sogou count must be 10/20/30/40/50
- user_id length must be between 6-128 characters if provided
- Search intent recognition increases response time but improves result relevance
Overview
Zhipu Search is a web search tool powered by the Zhipu AI API. It supports multiple engines from basic to advanced, including search_std, search_pro, search_pro_sogou, and search_pro_quark. It lets you filter results by time range or domain and to control how many results you receive and how detailed the results are.
How This Skill Works
It accepts a search_query and a chosen engine (search_std, search_pro, search_pro_sogou, search_pro_quark) and applies optional parameters such as search_intent, count, search_domain_filter, search_recency_filter, and content_size. The tool can be used via direct script calls or the OpenClaw Tool, and returns a structured response with identifiers and a list of results.
When to Use It
- Need the latest information from the web.
- Need to use a specific engine like Sogou, Quark, or Zhipu Search.
- Want to filter results by time range or domain to narrow sources.
- Need to control how many results you get and how detailed they are.
- Integrate with automation tools by leveraging optional params such as search_intent, request_id, or user_id.
Quick Start
- Step 1: Pick a query and a target engine, for example OpenClaw latest version with search_std.
- Step 2: Set optional parameters such as count, time filter, domain filter, and content_size as needed.
- Step 3: Run the search and review the structured results.
Best Practices
- Choose the engine that matches your content target (e.g., use search_pro_sogou for Chinese domestic content).
- Keep search_query within 70 characters to maximize relevance and performance.
- Use domain filters to whitelist trusted sources and avoid noise.
- Apply recency filters (oneDay, oneWeek, etc.) to prioritize fresh data.
- Respect rate limits and consider caching frequent queries for faster responses.
Example Use Cases
- Basic search for OpenClaw latest version using search_std with default settings.
- Advanced search for AI development trends with search_pro, count 20, search_recency_filter oneWeek.
- Chinese content lookup with search_pro_sogou, domain_filter example.cn, content_size high.
- Industry report search using search_pro_quark with domain filter www.example.org and recency noLimit.
- Direct script example: python scripts/zhipu_search.py --query AI trends --engine search_pro --count 10.
Frequently Asked Questions
Add this skill to your agents
