x-reader is a source-first Agent Skill, CLI and MCP server for reading URLs from X/Twitter, articles, video, podcasts, WeChat, Xiaohongshu, Telegram, RSS and the open web. It reports which source layers were actually retrieved (PASS / PARTIAL / FAIL / UNKNOWN), so an agent does not mistake a search snippet, a caption or a video description for the source itself.
MIT-licensed since its first release in February 2026; 988 GitHub stars and 97 forks as of November 2026. Works with Claude Code, Codex, Cursor and other MCP clients.
Fund this project