DeepSeek Harness Plugins · Client
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
A standalone Git status plugin for DSH Web, providing a side panel with a visual commit graph, uncommitted changes, and comprehensive branch operations, staging, and committing.
The dsh-cot-summerization plugin intercepts the model's raw chain of thought, rewrites it into a condensed version, enabling chain-of-thought hiding and optimization in open-source environments.
dsh-genome is an evolution engine for DSH capability transformation and assembly, managing skill/MCP/plugin repositories through a negative feedback loop for progressive stabilization.
A DeepSeek Harness plugin that enables users to interact with the Agent directly in personal WeChat via the official ClawBot functionality.
Share a DeepSeek Harness session over your LAN as a read-only, token-guarded snapshot.
The dsh-tesseract-ocr plugin enables text-only models to handle image attachments by performing local Tesseract OCR, sending only the recognized text to the model API, ensuring privacy.
dsh-task-notify is a DSH browser plugin that notifies you with a completion chime and tab-title alert when an agent turn finishes or a background job settles.
A Git graph view plugin for the DeepSeek Harness web interface that allows inspecting the workspace's Git history without creating conversation messages or trajectory entries.
dsh-turn-rail is a Codex-style auto-hiding turn rail for DeepSeek Harness Web featuring edge-triggered visibility, layered hit zones, search, and scrolling markers.
A DeepSeek Harness plugin for previewing Mermaid diagram images via local rendering in DSH Web, with fallback support.
dsh-drop-in plugin allows users to drag and drop files or folders from the system file manager into the DeepSeek Harness Web GUI, displaying them as a chip bar or inline reference bubbles with absolute paths for the agent to read.
A voice input plugin for DeepSeek Harness that provides a microphone button, integrates VocoType for offline speech recognition, and automatically inserts transcribed text into the chat input field.