A real-time Chinese web search skill for AI agents, leveraging the 360 search engine to deliver local news, policies, and market data.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install 360search
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install 360search using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is 360 Web Search?
360 Web Search is a technical integration designed to connect AI coding agents and LLMs directly to the 360 search engine. It serves as an optimized solution for handling real-time Chinese web searches, domestic news updates, and regional business directories. By integrating this skill, your agents can retrieve data that falls beyond their offline training cutoff limits.
As part of the Openclaw Skills ecosystem, this tool runs secure HTTPS requests directly to the 360 API and does not write to the host filesystem or execute local shell scripts. It delivers highly structured results complete with AI-generated semantic summaries, helping models understand and ingest search context without unnecessary token overhead.
360 Web Search Use Cases
- Querying breaking Chinese news, regional updates, and trending media content.
- Retrieving current profiles of domestic enterprises, policy changes, and financial market reports.
- Verifying facts or checking the status of products and public entities within the mainland Chinese web landscape.
- Grounding AI conversations and RAG pipelines using verified source URLs and precise publication times.
How 360 Web Search Works
- Intent Extraction: The AI agent evaluates the user prompt and determines if live Chinese web search or factual grounding is necessary.
- Credential Validation: The skill verifies the active presence of the
SEARCH_360_API_KEYenvironment variable. - Interface Mapping: Based on search intent, the skill selects the appropriate endpoint, defaulting to the robust
aiso-promixture search configuration. - Secure Request Dispatch: The skill generates a unique session ID (
sid) and transmits an HTTPS query toapi.360.cnalongside any customized parameters. - Response Extraction: The raw API response is parsed to capture document titles, original URLs, chronological timestamps (
page_time), and semantic fragments (summary_ai). - Context Synthesis: Cleaned search data is formatted and fed back into the agent context, with official sites flagged accordingly to ensure correct source attribution.
360 Web Search Setup
1. Retrieve Your API Key
- Log in to the 360 AI Platform.
- Navigate to Open Platform -> API Key Management.
- Create an application and copy the generated API Key.
2. Configure Your Environment Variables
Set the SEARCH_360_API_KEY variable on your host environment. Replace your_actual_key with the key obtained from the platform:
- Linux (bash):
echo 'export SEARCH_360_API_KEY="your_actual_key"' >> ~/.bashrc && source ~/.bashrc
- macOS (zsh):
echo 'export SEARCH_360_API_KEY="your_actual_key"' >> ~/.zshrc && source ~/.zshrc
- Windows (PowerShell):
setx SEARCH_360_API_KEY "your_actual_key"
3. Restart Application
Fully restart your AI terminal or developer environment to register the new environment variable. The skill will automatically detect the key upon the next search request.
360 Web Search Data Schema & Taxonomy
API Parameters (Web Search - /v2/mwebsearch)
| Parameter | Type | Required | Description |
|---|---|---|---|
q |
string | Yes | The search query string. Special characters like + must be URL-encoded (e.g., %2B). |
ref_prom |
string | Yes | The specific query engine configuration (e.g., aiso-pro or aiso-sr). |
sid |
string | Yes | A unique request UUID/session identifier generated dynamically. |
count |
integer | No | Maximum search results to return (range 1–20, default 10). |
summary_len |
integer | No | Target length for the AI semantic summary (range 120–3000, default 500). |
exclude_aigc |
string | No | Set to true to remove AI-generated content from the final results to avoid hallucinations. |
Response Payload Structure (Common Top-Level)
| Field | Type | Description |
|---|---|---|
errno |
integer | Error status code. 0 indicates successful operation. |
message |
string | Detailed explanation of error if errno is non-zero. |
count |
integer | Total count of results returned. |
items |
array | The core array containing specific search hits and metadata. |
Search Item Structure
| Field | Type | Description |
|---|---|---|
title |
string | The document or webpage title. |
url |
string | The target source destination URL (mandatory output). |
page_time |
string | Formatted publication or crawling timestamp (e.g., 2025-02-12 12:12:12). |
summary_ai |
string | Semantically extracted contextual summary (preferred field for LLM processing). |
official_site |
integer | Set to 1 if the webpage belongs to an officially verified entity. |
360 Web Search Advanced Features
- Diverse Search Interfaces: Supports specialized configurations like
aiso-pro(the default advanced knowledge query),aiso-max(multi-query rewrite),aiso-news(news-oriented filtering), andaiso-km-e2(boutique knowledge base targeting). - Comprehensive AI Search Engine: Integrates a POST-based AI Search endpoint (
/v1/search/aisearch) providing conversational summaries, complete references, and raw streams compatibility. - Image-to-Image and Text-to-Image Searching: Employs vertical SaaS pathways (
360so-v-ikand360so-v-ig) to conduct keyword-based image searches or reverse image similarity lookups. - AIGC Suppression Filtering: Allows setting
exclude_aigc=trueto isolate search returns to human-authored materials, decreasing recursive loop patterns in agent processing. - Authority-Based Weights: Implements system and manual credibility thresholds (
site_weight_sysandsite_weight_manual) ranging from 0 to 8 to omit low-quality websites dynamically.
SKILL.md
Loading
Related Openclaw Skills
Sector Opening Note
Sector Opening Note uses Openclaw Skills to research current HK tech headlines and turn them into a concise daily pre-market brief.
Compliance Policy Validator
An Openclaw Skills compliance scanner for Kubernetes and Docker that flags SOC2, HIPAA, NIST, and custom policy violations with fix hints.
Doubao Watermark Removal
Openclaw Skills automates batch Doubao watermark removal for AI images using OpenCV inpainting and configurable rectangular masks.
Agentic Framework Auditor
Agentic Framework Auditor in Openclaw Skills scans agentic projects for prompt drift, instruction conflicts, injection exposure, and production-readiness gaps.
Avatar Content Engine
A structured AI agent pipeline for generating emotionally-driven, high-velocity short-form video content featuring consistent virtual avatars.
Taobao Product Reviews
A browser automation skill designed to extract customer reviews, buyer options, and photo attachments from Taobao and Tmall product pages.
Featured*
