XNews

x-reader Open-Source MCP Server Reads 10+ Platforms with Evidence-Aware Output

x-reader Open-Source MCP Server Reads 10+ Platforms with Evidence-Aware Output

x-reader is an open-source content reader MCP server for AI agents that supports X, YouTube, Bilibili, WeChat, Xiaohongshu, podcasts and more than 10 platforms, normalizing content into structured source material. It reports which layers it actually read, marking partial reads and stating failures rather than presenting incomplete summaries as complete. A test with a tweet containing a nearly three-minute video captured the post text, the video and its Chinese subtitles.

Event version 1.

Key insights

Evidence-aware reading

x-reader distinguishes source layers: a search snippet is not the original text, a tweet caption is not the attached video, and a video description is not what was said. It marks missing layers as partially read and reports failures directly.

Multi-platform coverage

The tool reads X/Twitter, articles, YouTube, Bilibili, podcasts, WeChat, Xiaohongshu, Telegram, RSS and the open web, normalizing results into structured source content for research, summarization and automation.

Video and subtitle handling

Bilibili videos are processed via subtitles and audio transcription, and YouTube videos without subtitles fall back to speech-to-text. A test on a tweet with a nearly three-minute video captured the body text, video and Chinese subtitles.

Simple installation

A single command from the repository can be pasted into Claude Code, Codex or Cursor; the agent installs the tool, runs a self-check and reads the first link.

Source posts