dsh-zhihu-dashboard

Zhihu content is scattered across hot lists, follow feeds, hundreds of answers under a single question, and author pages that update at unpredictable intervals. Keeping a systematic watch on a keyword, an author, or a question thread usually means repeatedly refreshing tabs and manually diffing old versus new posts, leaving little room to hand that stream to an AI for synthesis. dsh-zhihu-dashboard folds the entire workflow into a sidebar panel inside DeepSeek Harness: the hot list shows trend arrows (new / up / down) computed from successive snapshots, the feed aggregates your favorites, your own creations, and the people you follow, and post tracking monitors questions, keywords, and people in parallel. New content is flagged NEW and surfaced inline under each tracked item, while a background checker running in the DSH top window fires system notifications and lights the sidebar unread badge even when the panel is closed.

Beyond the panel, the plugin registers seven native agent tools—Zhihu search, hot list, Zhida retrieval-augmented answers, whole-web search with a recency window, followees list, my creations, and favorites—so the model can pull Zhihu data directly inside a conversation. The smart-brief engine auto-routes candidates into templates based on the tracked object type and generates a polished brief with whatever DSH provider model is currently active. The opportunity report aggregates candidates from hot-list and feed caches, tracked NEW items, the unread queue, the content workspace, and research projects, scores them locally, then asks the model for a development-opportunity write-up. Every list can be exported as Markdown or CSV in one click, ready to paste into an agent conversation for deeper analysis. A clean reading mode, local keyword/author/regex block rules, and a content workspace (Later / Analyze / Done) round out the daily-driver experience.

This plugin fits developers, product managers, and researchers who browse Zhihu daily and want to fold that content into an AI-assisted workflow. It is especially useful for teams that need to track industry keywords, competitive moves, or academic question threads continuously, and prefer to synthesize findings, generate briefs, and evaluate opportunities through natural-language conversations with their agent.