zhipu-web-search

# Zhipu Web Search Web search via Zhipu AI's dedicated API (`/paas/v4/web_search`), refactored to use lightweight `cURL` instead of Python or `jq`. It defaults to using the `search_pro_quark` engine with 20 results. ## Quick Start ### Basic cURL Usage ```bash curl --request POST \ --url https://open.bigmodel.cn/api/paas/v4/web_search \ --header "Authorization: Bearer $ZHIPU_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ "search_query": "OpenClaw framework", "search_engine": "search_pro_quark", "search_intent": false, "count": 20 }' ``` ### Script Usage A wrapper shell script is provided for convenience. ```bash # Basic Search (defaults to search_pro_quark and 20 results) bash scripts/zhipu_search.sh --query "AI development trends" # Advanced Search bash scripts/zhipu_search.sh \ --query "latest open source LLMs" \ --engine "search_pro_sogou" \ --count 50 \ --intent \ --recency "oneWeek" ``` ## API Parameter Reference | Parameter | Type | Required | Default | Description | |-----------|------|----------|---------|-------------| | `search_query` | string | ✅ | - | Search content, recommended ≤70 chars | | `search_engine` | enum | ✅ | `search_pro_quark` | `search_std` / `search_pro` / `search_pro_sogou` / `search_pro_quark` | | `search_intent` | boolean | - | `false` | Enable search intent recognition | | `count` | integer | - | `20` | Result count, range 1-50 | | `search_domain_filter` | string | - | - | Whitelist domain filter | | `search_recency_filter` | enum | - | `noLimit` | `oneDay` / `oneWeek` / `oneMonth` / `oneYear` / `noLimit` | | `content_size` | enum | - | `medium` | `medium` (summary) / `high` (detailed) | ## Search Engine Selection Guide | Engine | Use Case | |--------|----------| | `search_pro_quark` | Quark search, tailored for specific advanced scenarios (**Default**) | | `search_std` | Basic search, regular Q&A | | `search_pro` | Advanced search, need more accurate results | | `search_pro_sogou` | Sogou search, China domestic content | ## Response Structure The API returns JSON directly. ```json { "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_KEY` must be configured. - `curl` command must be available in your system path.