dsh-lmstudio-router
Run the following command in DeepSeek Harness:
dsh plugin install xiaoyizhuang03-droid/dsh-lmstudio-router
Paste the following prompt into your AI chat to install this plugin:
Install the plugin in DeepSeek Harness by running the install command; the full source repository is available at https://github.com/xiaoyizhuang03-droid/dsh-lmstudio-router
About this plugin
dsh web handles a steady stream of routine but token-heavy work behind the scenes — session titling, context compaction, bulk text reshaping. Each call is individually cheap, yet together they add up on a remote API bill. dsh-lmstudio-router identifies exactly those low-priority, high-cost requests and redirects them to a local LM Studio model, leaving the remote API free for tasks that genuinely need cloud-scale reasoning.
Under the hood the plugin speaks to LM Studio over its OpenAI-compatible /v1 endpoint, auto-launches the server when it is not already running, and trims every local request to fit within the context window of the target model (default 8,192 tokens) so nothing overflows. It is a dependency-free host plugin with no extra services or sidecars; users can also prepend [local] or [remote] to any message to force routing on the fly.
If you already keep LM Studio on a workstation and want to peel everyday dsh web chores off your remote API to cut both cost and latency, this is a lightweight entry point. One config line, a restart, and the routing is live — no model swaps, no additional infrastructure.
Use Cases
- Route dsh web chores such as titling, compaction, and bulk text reshaping to a local LM Studio automatically
- Cut remote API token spend by offloading low-priority requests to a locally running LM Studio instance
- Force per-message routing with [local] or [remote] prefixes to override the default policy
Best For
- dsh web users who already run LM Studio on their workstation
- Developers wanting to trim their remote API monthly bill
- Users who prefer zero-dependency, plug-and-play dsh host plugins
Related Plugins
A unified suite combining hot runtime injection, task-aware thinking-mode routing, and a graded session protocol with red-team gates to sustain model diligence across long-horizon inference.
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.