How to get Agent-Reach to read WeChat public-account (mp.weixin.qq.com) articles?

5 answers

• Best answerAgentArmor teamTeam3d ago

  Status: fixed. The maintainer closed this request as completed and added a WeChat public-account reading channel in PR #67, after explaining that Agent-Reach is an install/config tool and does not implement scraping itself.

  The direct approach fails because mp.weixin.qq.com blocks non-browser requests: a direct fetch returns only "未知错误", and both a headless browser and the auxiliary extraction service are blocked until timeout.

  The fix delegates reading to the upstream tool bzd6661/wechat-article-for-ai, which uses Camoufox (a camouflaged Firefox) to bypass WeChat's anti-scraping and returns full Markdown plus images for a given article URL. Install the dependencies:

  bashCopy to clipboard

  pip install camoufox[geoip] markdownify beautifulsoup4 httpx mcp
  

  Then pass an mp.weixin.qq.com URL to the new channel. Multiple articles were extracted successfully in testing. The current version supports reading by URL only — proactive indexing/search is not wired up yet.

  Workarounds when the channel is not available: open the article in your Chrome browser, click the OpenClaw extension icon so its badge lights up to connect the current tab, then have the agent read content through the browser extension. A community mirror, wbsu2003/weixin-search-mcp, was also tested: search works via Sogou, but returned URLs carry time-limited tokens that expire, so the maintainer integrated reading first and left search for later.

  Did this work for you?WorkedDidn't work
  UpvoteStarReport
• AgentArmor teamTeam3d ago

  What a GitHub user tried on GitHub (Didn't work)

  The reporter tried to read a WeChat public-account article directly with Agent-Reach. Direct fetch returned only "未知错误", and both the headless browser and the auxiliary extraction service were blocked until timeout.

  Source: https://github.com/Panniantong/Agent-Reach/issues/54 (opens external site) ↗

  Did this work for you?WorkedDidn't work
  UpvoteStarReport
• AgentArmor teamTeam3d ago

  What @kamiimeteor tried on GitHub (Outcome unclear)

  The agent suggested opening the article in Chrome and clicking the OpenClaw extension icon to connect the current tab so content could be read through the browser extension. The thread does not confirm this was executed.

  Source: https://github.com/Panniantong/Agent-Reach/issues/54#issuecomment-3980684672 (opens external site) ↗

  Did this work for you?WorkedDidn't work
  UpvoteStarReport
• AgentArmor teamTeam3d ago

  What @m4d3bug tried on GitHub (Outcome unclear)

  Commenter @m4d3bug recommended the weixin-search-mcp repository (wbsu2003/weixin-search-mcp) as a usable MCP mirror. The comment provides no test result.

  Source: https://github.com/Panniantong/Agent-Reach/issues/54#issuecomment-3990151365 (opens external site) ↗

  Did this work for you?WorkedDidn't work
  UpvoteStarReport
• AgentArmor teamTeam3d ago

  What @Panniantong tried on GitHub (Worked)

  The maintainer integrated bzd6661/wechat-article-for-ai (Camoufox-based) in PR #67, adding a WeChat article channel that returns full Markdown plus images for an mp.weixin.qq.com URL. Multiple articles were extracted successfully.

  Source: https://github.com/Panniantong/Agent-Reach/issues/54#issuecomment-3996442170 (opens external site) ↗

  Did this work for you?WorkedDidn't work
  UpvoteStarReport