DeepSeek Harness Plugins
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
The plugin warms the projection cache at startup, enabling the sidebar to display each session's theme directly without needing to open the session first.
Sidor_UI is a starry-themed skin for DeepSeek Harness Web GUI with opening animations, persistent starry elements, balance badges, and low-balance warnings for fun dress-up.
dsh-plugin-rollout-scout is a rollout scout plugin that analyzes streaming chain-of-thought output and paragraph openings, using heuristic scoring based on phrasing to identify new chat models in gray-release deployments.
dsh-plugin-chrome opens a visible Chrome window per session for agent control via the chrome_* tool suite, streams live view to the Web GUI, and allows manual takeover.
dsh-session-nexus is a community plugin for DeepSeek Harness offering a complete session toolkit with lifecycle control, cross-session messaging, and address management through commands and model tools.
Provides a persistent usage ledger and visual dashboard for DeepSeek Harness Web profile, featuring session history backfill, flexible filtering, and precise token statistics.
A DSH WeChat bridge plugin that enables direct conversation with a local AI agent from WeChat, supporting multi-user interactions, file exchange, and proactive notifications.
A searchable, sortable model picker for the DeepSeek Harness Web UI, featuring a table-style interface with search, sorting, favorites, local filtering, and more.
Displays the current provider's balance or usage in the status bar below the input field in DeepSeek Harness, with real-time switching and caching across multiple providers.
This plugin integrates xquant's quantitative research MCP into DeepSeek Harness, allowing users to query factors, strategy logic, and historical data for quantitative analysis.
meow-vision offers image recognition for text models and UI preview screenshots for multimodal models, supporting component design iteration.
dsh-rewind folds exploration since the last checkpoint into an auto-generated report, optimizing model context.