dsh-bash-rtk
Run the following command in DeepSeek Harness:
dsh plugin install DeepTrial/dsh-bash-rtk
Paste the following prompt into your AI chat to install this plugin:
To install this plugin in DeepSeek Harness, run dsh plugin install DeepTrial/dsh-bash-rtk; the full source repository is at https://github.com/DeepTrial/dsh-bash-rtk.
About this plugin
In LLM agent development, verbose tool outputs such as git logs or build trails consume excessive tokens, increasing costs and reducing efficiency. The dsh-bash-rtk plugin addresses this by automatically routing eligible shell commands through rtk (Rust Token Killer) to compress outputs and save tokens without altering execution semantics. Its core capability lies in intelligent filtering: it only handles simple, whitelisted commands while bypassing complex shell structures with pipes or redirects, and gracefully falls back to native execution if rtk is unavailable. This plugin is ideal for developers using DeepSeek Harness for model inference, especially those seeking to optimize token usage while maintaining stable shell execution.
Use Cases
- Automatically compress output when LLM agents run git or cargo commands
- Process build logs or test outputs to reduce token usage
- Optimize tool responses while preserving execution semantics
Best For
- LLM agent developers using DeepSeek Harness
- Data scientists focused on inference cost optimization
- AI engineers needing to streamline tool outputs
Related Plugins
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.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.