AI Agent Hub
Back to plugins
🧰

dsh-web-search-anysearch

Web Tools Updated 2026.09.03

Run the following command in DeepSeek Harness:

dsh plugin install JPA957/dsh-web-search-anysearch

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

Run dsh plugin install JPA957/dsh-web-search-anysearch in DeepSeek Harness to install the plugin; full source repository: https://github.com/JPA957/dsh-web-search-anysearch

About this plugin

DeepSeek Harness exposes a pluggable web-search seam, but wiring up AnySearch often meant hand-writing multiple patch sections and manually registering the provider, with configuration scattered across files. dsh-web-search-anysearch packages all of that into an install-ready plugin: a dedicated AnySearch tab in the web app Settings lets you enter the API key, endpoint base, and max result count with click-to-save persistence, eliminating patch-file editing entirely.

The core is a robust key-pool scheduler. Comma-separated credentials form a round-robin pool; on each successful search the cursor advances. When a key hits a 429 rate limit (honoring the Retry-After header), quota exhaustion, or a 401/403 auth rejection, that key goes on cooldown (60 s for rate limits, 10 min for auth and quota) and the current search immediately retries the next key, up to six attempts, transparent to the caller. Keys enter only through the credentials domain; the browser sees only a configured or unset indicator, and diagnostic logs mask every key to its first ten characters. The result pipeline drops entries without a URL, collapses duplicate links, and prefers the content field over snippet.

Ideal for teams that already rely on AnySearch as their search backend, manage multiple API keys requiring rotation, and prefer a visual settings page over scattered patch-file edits. Migrating from a hand-mounted copy? Keep the searchProvider: anysearch pin, remove the old insert line, and the bundle's own registration layer takes over seamlessly.

Use Cases

  • Connecting AnySearch as the default web-search backend for DeepSeek Harness
  • Pooling multiple API keys with round-robin rotation and automatic cooldown failover on 429 or 401/403
  • Configuring key, endpoint, and result count in the Settings tab without any patch-file editing

Best For

  • Teams already running AnySearch as their production search backend
  • Developers who need multi-key rotation to avoid single-key rate-limit outages
  • Application developers who prefer a visual settings page over scattered patch-file edits