AI Agent Hub
Back to plugins
dsh-catgirl-plugin preview

dsh-catgirl-plugin

Model Inference Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install Freakz2z/dsh-catgirl-plugin

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

In DeepSeek Harness, install this plugin by running `dsh plugin install Freakz2z/dsh-catgirl-plugin`,with the source code available at https://github.com/Freakz2z/dsh-catgirl-plugin。

About this plugin

Traditional persona injection in LLM applications often inflates token usage and costs per request while cluttering model outputs. This plugin addresses that with Persona Virtualization: it offloads the catgirl flavor to local rendering, keeping the model context lean and token-efficient. Key features include aggressive tool schema pruning and an on-demand unlock system, which real-world tests show reduce input tokens by up to 67% without degrading quality across tasks like coding, web search, and pure Q&A. It's ideal for developers, AI enthusiasts, and teams looking to optimize LLM interactions in DeepSeek Harness—whether for cost savings, faster responses, or scalable deployments—without compromising on functionality or output flair.

Screenshots

Use Cases

  • Routine development tasks requiring reduced LLM interaction costs.
  • Complex workflows involving coding, file search, and multi-tool coordination.
  • Adding fun, personalized interactions while maintaining intelligent output.

Best For

  • Developers and power users of DeepSeek Harness.
  • AI researchers focused on LLM cost optimization and efficiency.
  • Product designers seeking a balance between intelligence and playfulness.