BigModel Web Search Fallback for Openclaw

A non-invasive web search fallback for AI agents that leverages Zhipu BigModel and GLM engines to provide structured or summarized results.

gxhdky2345
v1.0.0
Mar 21, 2026
0
738
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install bigmodel-web-search-fallback

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install bigmodel-web-search-fallback 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 BigModel Web Search Fallback?

The BigModel Web Search Fallback is a specialized tool designed to augment Openclaw Skills by providing a reliable search alternative when default routes fail. It acts as a local wrapper script that interfaces directly with Zhipu Web Search APIs, offering two primary execution modes: raw for structured data and chat for synthesized summaries. This skill ensures that your AI agents maintain high availability and access to diverse search engines like Sogou and Quark without requiring core modifications to the agent architecture.

BigModel Web Search Fallback Use Cases

  • When the built-in web search route is unavailable, failing, or returning low-quality results.
  • Scenarios requiring high-quality Chinese web search data directly from Zhipu or GLM sources.
  • Comparative research tasks that benefit from results across different providers like Sogou or Quark.
  • Situations where the AI agent needs raw JSON structured results for programmatic data processing and custom citations.

How BigModel Web Search Fallback Works

  1. The AI agent identifies a need for web search and determines if the BigModel fallback is the preferred route.
  2. The system selects between the raw execution mode for structured data or chat mode for a synthesized GLM answer.
  3. An appropriate search engine is chosen based on the query type, such as search_std for speed or search_pro for depth.
  4. The wrapper script executes the API call using the configured environment credentials.
  5. The script returns formatted JSON to stdout, which the agent parses to provide the final user response or to drive further workflows.

BigModel Web Search Fallback Setup

To integrate this search capability into your Openclaw Skills, ensure you have an active API key from BigModel. Set one of the following environment variables on your host:

export ZAI_API_KEY='your_api_key_here'
# Alternatively use ZHIPUAI_API_KEY or BIGMODEL_API_KEY

Test the installation by running the script from the command line:

python scripts/zhipu_web_search.py raw --query "OpenClaw search" --engine search_std --count 5 --pretty

BigModel Web Search Fallback Data Schema & Taxonomy

The skill returns data in a structured JSON format, making it easy for agents to parse and use. The schema supports several input flags and output attributes:

Feature Details
Execution Modes raw (structured links) or chat (synthesized text).
Engine Types search_std, search_pro, search_pro_sogou, search_pro_quark.
Filters Supports --domain-filter and --recency (time-based) constraints.
Output Fields Includes title, link, summary, media, and date for raw results.

BigModel Web Search Fallback Advanced Features

  • Multi-Engine Routing: Ability to specifically target Sogou or Quark backed routes for specialized regional retrieval.
  • Non-Invasive Integration: Enhance Openclaw Skills without modifying the internal core code of the AI coding agent.
  • Dynamic Context Control: Adjust content-size and count flags to manage the volume of data returned to the agent's context window.
  • High-Quality Synthesis: Utilize the chat mode to leverage GLM's internal reasoning for one-step search and summarization tasks.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*