dsh-presets-zh
Run the following command in DeepSeek Harness:
dsh plugin install GongYuanCaiJi/dsh-presets-zh
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install GongYuanCaiJi/dsh-presets-zh inside DeepSeek Harness to install this plugin. The source repository is available at https://github.com/GongYuanCaiJi/dsh-presets-zh . Open a new session after installation for it to take effect.
About this plugin
DeepSeek is a Chinese-native model, yet dsh ships its four agent presets (standard, cordis, code, minimal) with English-only system prompts. When a Chinese-speaking user asks a question, the model first translates internally to English, reasons in English, then translates back to Chinese for output—an extra hop that blurs tone and nuance. dsh-presets-zh translates all four preset system prompts into Chinese and loads them automatically by host locale: Chinese environments get the -zh variants, and switching back to English restores the originals instantly.
The translation is faithful and minimal: only user-facing narrative and instruction text is changed. Tool names, plugin names, !!js expressions, and other structural lines are preserved verbatim. Edited user presets are never overwritten; the plugin rewrites default files only on suite upgrade when the user has not modified them. The two composition skills bundled with cordis-zh remain in English, as they are outside the scope of this port.
Built for Chinese-speaking DeepSeek users who want the reasoning chain and the output in the same language, eliminating the extra translation layer. Upstream is deepseek-ai/deepseek-harness (MIT); this port is licensed under MIT with dual copyright notices in LICENSE and full attribution in THIRD_PARTY_NOTICES.md.
Use Cases
- Conversing with DeepSeek in Chinese daily so the reasoning chain and output stay in the same language
- Auto-loading the matching *-zh preset variant when toggling between Chinese and English locales
- Composing in Chinese with cordis-zh to avoid the extra translation hop injected by English system prompts
Best For
- Chinese-speaking users who interact with DeepSeek in Chinese as their primary language
- Engineers and prompt crafters who want to remove the extra translation layer from inference chains
- Content creators who fluidly switch between multiple languages inside dsh
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.