dsh-custom-context-injector
Run the following command in DeepSeek Harness:
dsh plugin install LiKanGame/dsh-custom-context-injector
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install LiKanGame/dsh-custom-context-injector inside DeepSeek Harness to install; the source repository is https://github.com/LiKanGame/dsh-custom-context-injector .
About this plugin
Pasting your role definition, output preferences, or project-specific terminology before every model call is tedious and easy to forget. dsh-custom-context-injector adds a dedicated "Custom Context" page to the DeepSeek Harness settings panel. Toggle the injection switch on, type your context, hit Save, and from then on the text is automatically injected as a systemPrompt section into every model assembly — in the current session and in every new conversation you open afterwards.
Persistence uses a two-tier strategy: an in-memory cache serves synchronous injection at call time, while a local JSON file (.dsh-custom-context.json in the DSH host process working directory) is written on save and reloaded on restart. If disk writing is unavailable the plugin gracefully degrades to an in-memory-only mode and surfaces a notice on the settings page. The injection scope is limited to the realm that loaded the plugin and affects subsequent prompt assembly; it does not retroactively rewrite historical sessions.
This plugin is ideal for users who want a persistent persona, fixed output style, project terminology, or team-level conventions attached to every Harness conversation. Note that it currently ships as a dynamic Cordis Package relying on services injected by the DSH host and is not yet published as a standalone npm package; a formal release is planned once it migrates to a persistent settings namespace or a composition directory.
Use Cases
- Maintain a fixed persona or role definition across every turn without repeated pasting
- Enforce consistent output format, tone, or style requirements without manual reminders each time
- Inject project terminology, team conventions, or domain constraints to reduce corrections
Best For
- Individual users who need a consistent persona or output style in every conversation
- Team members who require unified dialogue conventions and terminology
- Engineers developing DSH plugins on the Cordis dynamic runtime
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.