AI Agent Hub
Back to plugins
🧰

dsh-tinyfish-search

Web Tools Updated 2026.09.04

Run the following command in DeepSeek Harness:

dsh plugin install maxwell-feng/dsh-tinyfish-search

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

Run the install command in the DeepSeek Harness terminal; the source repository is https://github.com/maxwell-feng/dsh-tinyfish-search — once installed, the built-in web_search tool is automatically routed through the TinyFish Search provider.

About this plugin

The built-in web_search tool in DeepSeek Harness routes through an Anthropic-compatible endpoint by default, meaning every search consumes a full model turn adding both latency and an extra slice of token cost. dsh-tinyfish-search registers a stable tinyfish provider on the ctx.web capability seam so each web_search call becomes a single GET request to the TinyFish Search API, returning structured titles, snippets, links, and dates without ever touching an LLM—faster and cheaper.

Once installed, the plugin takes over the built-in web_search automatically with zero code changes to your sessions. You can configure the API key, geo location, and search language via environment variables or profile settings; saved edits take effect on the very next search without a restart. Uninstalling cleanly restores every seam row to its original defaults.

Ideal for developers working with the dsh web profile who want zero model cost per search, lower response times, or the ability to pin results to a specific region or language—dsh-tinyfish-search is a lightweight, drop-in alternative that just works.

Use Cases

  • Reduce per-search model token cost to zero
  • Pin search results to a specific region or language
  • Replace the default DeepSeek backend for faster responses

Best For

  • Developers building on the dsh web profile
  • Conversational apps sensitive to search latency
  • Teams that want fine-grained control over token budgets