AI Agent Hub
Back to plugins
🧰

dsh-ventus-search

Web Tools Updated 2026.08.24

Run the following command in DeepSeek Harness:

dsh plugin install meto-ventus/dsh-ventus-search

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install meto-ventus/dsh-ventus-search in the DeepSeek Harness terminal to install this plugin; the repository is available at https://github.com/meto-ventus/dsh-ventus-search . Restart dsh after installation to load the new bundle.

About this plugin

Running web searches inside DeepSeek Harness with a single engine often means dealing with timeouts, anti-bot redirects, ad clutter, and the absence of any unified ranking or fault tolerance. dsh-ventus-search registers Bing, 360, and Bilibili as a group of concurrent providers under ctx.web. Results are scored by title hit, snippet hit, and rank weight, then normalized, deduplicated, redirect-decoded, and capped per domain so a single query returns clean, clickable sources.

The resilience chain is thorough: a failing engine is isolated automatically, a backup query retries, an overall timeout returns whatever partial results exist, and if every engine fails the plugin returns an empty array instead of throwing. A query-level LRU cache (capacity 100, TTL 300 s) combined with AbortSignal propagation keeps repeat requests nearly free. The fetch side includes an ad and tracker domain blacklist with subdomain matching, a Readability-style content extractor, mirror-domain fallback, and a size cap, so consumers get readable article text rather than raw HTML.

It is aimed at developers who need frequent Chinese and English web lookups inside DSH conversations without hand-wiring each search engine API, as well as teams building agents that require dependable web_search and web_fetch primitives with observable per-engine health, a visual settings card, and graceful degradation out of the box.

Use Cases

  • Search Chinese and English web content in DSH conversations with ad-free article extraction
  • Provide reliable and observable web_search and web_fetch primitives to your AI agent
  • Concurrent multi-engine search that degrades gracefully when any single engine fails

Best For

  • Developers who need frequent web lookups inside a DSH workflow
  • Platform engineers building stable search and fetch pipelines for AI agents
  • Admins who want per-engine health indicators and independent on-off toggles