AI Agent Hub
Back to plugins
um-dsh-websearch preview

um-dsh-websearch

Web Tools Updated 2026.09.04

Run the following command in DeepSeek Harness:

dsh plugin install UnforgetMemory/um-dsh-websearch

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

Run dsh plugin install UnforgetMemory/um-dsh-websearch in DeepSeek Harness to install; source at https://github.com/UnforgetMemory/um-dsh-websearch

About this plugin

When a web search backend goes down, runs out of quota, or rejects a request, a single-provider setup in DeepSeek Harness leaves you without any search at all. um-dsh-websearch mounts both Exa and Parallel under one umbrella provider, turning primary-to-failover switching from a code change into a one-line config, so a single point of failure no longer stalls your connected experience.

Inside each backend, two channels work as mutual backups: authenticated REST and keyless anonymous MCP. If one channel hits a 401, 429, or 5xx, the other takes over automatically; cross-backend fallback works the same way, so a total rejection on the primary seamlessly routes to the secondary. In keyless mode, public MCP endpoints return results without any API key, making quick validation and lightweight deployment painless. The global toggle ships disabled, flips on with one click in settings (hot, no restart), and every field change takes effect on the next search.

If you want a redundancy layer for DeepSeek web search—multi-backend failover, zero-key lightweight access, or a bilingual settings card where priority and fallback policy are visible at a glance—this plugin consolidates scattered configuration into a single card. Install it and it takes over the default search; uninstall it and the original backend is restored, with no manual profile edits required.

Screenshots

Use Cases

  • Auto-fallback to Parallel when Exa quota expires or rate-limits
  • Validate search instantly via anonymous MCP endpoints with zero keys
  • Toggle search priority and backend state from a bilingual settings card

Best For

  • DeepSeek production deployments needing multi-backend search redundancy
  • Individuals or small teams wanting zero-key lightweight search access
  • DevOps engineers managing search policies across environments