Zhihu Search and Content Organizer

Problem

Zhihu content is spread across answers, articles, and question pages. When an agent scrapes HTML or consumes raw API fields directly, it often has to handle summary length, author metadata, vote/comment counts, and inconsistent error bodies. This skill narrows that workflow to a clearer search step: pass a query and an optional count, then receive normalized items[] results.

How It Works

It calls the Zhihu Open Platform endpoint GET /api/v1/content/zhihu_search using the zhihu_search API and authenticates with ZHIHU_ACCESS_SECRET. At runtime, it reshapes the response into a more stable JSON structure containing code, message, item_count, and item fields such as title, summary, url, author_name, vote_up_count, comment_count, and edit_time. If the request returns a non-2xx HTTP status, the error object includes a dynamic error message and, when available, the original response body, which helps diagnose gateway, authentication, or parameter issues.

Boundaries and Notes

This skill is useful for searching Zhihu content, summarizing results, and preparing material for follow-up Q&A or note generation. It does not replace the full Zhihu platform experience, such as logged-in interactions, private messaging, social graph features, or APIs outside the Open Platform scope. Callers must provide their own Open Platform Access Secret and respect platform quotas. count is limited to 1-10, so it fits small-batch retrieval rather than bulk crawling.