AI Agent Hub
Back to plugins
🤖

dsh-switch

Model Inference Updated 2026.08.15

Run the following command in DeepSeek Harness:

dsh plugin install dongsheng123132/dsh-switch

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

To install the dsh-switch plugin in DeepSeek Harness, use the command: dsh plugin install dongsheng123132/dsh-switch, or obtain the source code from the GitHub repository at https://github.com/dongsheng123132/dsh-switch.

About this plugin

In the DeepSeek Harness ecosystem, managing model routes can become complex as the number of providers and paths increases. Users often struggle to reliably verify and monitor model selections, leading to potential errors in routing configuration. dsh-switch addresses these challenges as an evidence-first model control plane.

dsh-switch offers two key capabilities. First, via its DSH bundle, it exposes tools that inspect registered provider routes, validate specific provider and model combinations, and set defaults for new sessions. This ensures accuracy and consistency in model selection. Second, the standalone CLI probes OpenAI-compatible endpoints, reporting reachability, latency, model counts, and credential configuration status without consuming inference tokens, enabling cost-effective monitoring.

This plugin is ideal for DeepSeek Harness developers and operations teams. If you manage multiple model providers, need to validate routing configurations, or want to monitor endpoint health in real-time, dsh-switch provides a transparent and controllable solution. It doesn't interfere with existing request flows but acts as an observable control plane, enhancing your model management capabilities.

Use Cases

  • Verify route configurations for model providers.
  • Probe reachability and performance metrics of OpenAI-compatible endpoints.
  • Set and switch default models for new sessions.

Best For

  • DeepSeek Harness developers managing model routes.
  • Operations teams monitoring API endpoint availability.
  • AI project leaders ensuring accurate and consistent model selection.