Zh Mcp Server

Overview

The zh_mcp_server is a Model Context Protocol (MCP) server specifically engineered to bridge large language models with the Zhihu platform. By exposing a standardized MCP interface, it allows AI assistants such as Claude or custom agents to generate and publish content directly on Zhihu without manual intervention. This solves the friction that developers face when integrating AI-generated text into social‑media workflows, enabling a fully automated publishing pipeline from model inference to live article posting.

At its core, the server orchestrates three essential steps: (1) authenticating a user through Chrome‑based cookie capture, (2) invoking a language model to produce article drafts, and (3) submitting those drafts via the Zhihu web interface using Selenium‑controlled automation. The MCP contract exposes these operations as simple tools, so any client that understands MCP can request a “post‑to‑Zhihu” action and receive the resulting article URL or status. This abstraction hides browser automation complexities, allowing developers to focus on higher‑level logic such as content curation, scheduling, or multi‑platform syndication.

Key capabilities include:

• Cookie‑based authentication: Users run a small login helper that opens Chrome, logs in with phone verification, and stores the session cookie for future requests.
• Headless browser control: The server can operate in headless mode for production, yet offers a toggle to enable visual debugging when troubleshooting model interactions or UI changes.
• Encoding support: By default the server emits UTF‑8 output, but it also accepts an flag and environment variable to accommodate non‑UTF‑8 client environments, ensuring broad compatibility across languages.
• MCP tool registration: The service registers itself under the name , exposing a single command that client tools invoke via standard MCP JSON messages.

Typical use cases span content marketing, automated knowledge base creation, and social‑media management. A marketing team could configure a cron job that triggers the MCP server to generate weekly Zhihu articles based on trending topics, while a research lab could publish AI‑generated summaries of new papers directly to Zhihu for community dissemination. Because the server handles authentication and UI automation internally, teams can integrate it into existing AI workflows—whether in Python notebooks, Java Spring applications, or other MCP‑compatible frameworks—without writing custom web‑scraping code.

What sets zh_mcp_server apart is its focus on a single, high‑value platform (Zhihu) combined with the flexibility of MCP. Developers benefit from a plug‑and‑play tool that abstracts away browser automation, offers robust encoding handling, and can be easily swapped into any AI pipeline that already speaks MCP. This makes it an attractive component for developers looking to automate content creation and distribution on Chinese social‑media ecosystems.