dsh-universal-worldbook
Run the following command in DeepSeek Harness:
dsh plugin install TritiumWang/dsh-universal-worldbook
Paste the following prompt into your AI chat to install this plugin:
Run the command dsh plugin install TritiumWang/dsh-universal-worldbook in your terminal to install the plugin; the source repository is available at https://github.com/TritiumWang/dsh-universal-worldbook .
About this plugin
Roleplay, coding assistance, and research conversations share a common frustration: certain lore entries, behavioral constraints, or domain glossaries need to appear conditionally in the model context without ever being written into session history or leaking into subsequent turns. dsh-universal-worldbook solves exactly that. It rewrites the outbound request at the llm/stream layer, splicing prompt snippets in right before the user's latest message; once the request completes, the injection vanishes with no trace in the trajectory and no residue in the next context window.
The design pairs controlled injection with on-demand retrieval. Triggers support always-on insertion, keyword (regex-capable) matching, and an adjustable scan depth from 1 to 5, using the same filter logic as SillyTavern worldbooks. The main agent can invoke a built-in lore_lookup tool to query sub-entries by index, and results are stripped turn-by-turn so they never accumulate. Worldbooks are scoped per workspace, the entry UI mirrors SillyTavern with added drag-to-reorder and live hit preview, and all data is stored in SillyTavern-compatible JSON fully decoupled from plugin code so upgrades never wipe your content.
This plugin is ideal for DSH users who want a single structured worldbook to manage prompt fragments across every scenario. If you already rely on SillyTavern worldbooks and want finer trigger control, agent-driven entry lookup, or seamless reuse across workspaces, it can serve as your prompt management foundation.
Use Cases
- Conditionally injecting character settings and lore entries in roleplay
- Splicing domain glossaries and behavioral constraints into coding or research sessions
- Letting the agent look up worldbook sub-entries on demand without polluting later context
Best For
- Users who maintain SillyTavern worldbooks and want to reuse them in DSH
- Prompt managers needing fine-grained trigger control and agent-driven lore lookup
- DSH daily users with multi-workspace setups who require zero data loss on upgrades
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.