AI Agent Hub
Back to plugins
🧰

dsh-web-search-router

Web Tools Updated 2026.08.22

Run the following command in DeepSeek Harness:

dsh plugin install Kerberos255/dsh-web-search-router

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

Run dsh plugin install Kerberos255/dsh-web-search-router in your terminal to install the plugin; the source repository is available at https://github.com/Kerberos255/dsh-web-search-router

About this plugin

DeepSeek Harness ships with a single-engine web_search tool. When that engine times out, hits a rate limit, or exhausts its quota, the agent gets an empty result or a hard error with no recovery path. dsh-web-search-router inserts a priority-ordered multi-provider router behind the existing tool interface: the model keeps calling the same web_search tool while up to eight backends are tried in sequence until one returns a usable answer.

Eight providers ship out of the box—SearXNG, DeepSeek Search, Tavily, Brave Search, Exa, Firecrawl, Parallel, and DuckDuckGo—with unconfigured ones skipped automatically. Transient failures trigger a short cooldown; rate-limit and quota errors support longer custom cooldowns; and whether an empty result should fall through to the next provider is a simple per-router toggle. Non-secret settings are stored in DSH\u0027s native Settings namespace and take effect without a restart. API keys are managed exclusively through DSH Credentials, and router error messages are sanitized to redact URLs and token-like values before logging.

This plugin is for DeepSeek Harness users who want a resilient multi-engine search pipeline for their agents. As long as your agent preset exposes DSH\u0027s built-in web_search tool, you can reorder, enable, and disable providers from the Settings panel without changing any agent code.

Use Cases

  • A single search engine times out or hits a rate limit; the router automatically falls through to the next provider in priority order
  • Reorder, enable, or disable providers live from the Settings panel without touching any agent tool code
  • Wire up self-hosted SearXNG as the primary engine while keeping Tavily, Brave, or Exa as cloud backends

Best For

  • DeepSeek Harness developers who want more resilient web search for their agents
  • Teams hitting frequent timeouts, rate limits, or quota exhaustion on web_search calls
  • AI application engineers who prefer multi-engine or self-hosted search stacks