Leo introduces x-reader v0.3.1 and its source-first approach: read the original link and report what is still missing. The post describes a webpage fallback and distinguishes retrieved X text from attached media that has not been read; those update examples are the author’s reported results.
x-reader: read URLs with explicit evidence
Read original X posts, articles, video, podcasts and web sources before summarizing or analyzing them.
Source facts
- Repository
- runesleo/x-reader
- Last source activity
- October 14, 2026 at 17:27
- Detected SKILL.md language
- English
- Stars
- 988
- Forks
- 97
Examples
The author's first-success example documents an X post read through oEmbed, retrieval of its attached video and Chinese subtitles, and a broken-URL failure case. This is the author's reported run.
Prerequisites
Use an existing x-reader CLI when available. The Skill otherwise describes an isolated cache environment with an immutable CLI revision. Material audio or video needs a separate inspection or transcription path, such as yt-dlp with subtitles or audio transcription.
How to use
Give the agent the source URL and ask it to read the original content, consume any attached audio or video that matters to the answer, and report what it read and what remains missing. The source workflow then reads the complete CLI JSON before analysis.
Limitations
A text fetch does not establish that attached media was consumed. The Skill distinguishes PASS, PARTIAL, FAIL and UNKNOWN; login walls, deleted pages, empty responses and inaccessible attachments must remain explicit gaps. Fetched content is treated as untrusted source data, never as instructions to execute.
Install options
The review-first prompt is selected by default. You can switch to a direct command or download a local copy.
Review the source files
Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.
Showing SKILL.md
| name | x-reader |
|---|---|
| description | Read source URLs from X/Twitter, articles, YouTube, Bilibili, podcasts, WeChat, Xiaohongshu, Telegram, RSS, and the open web. Use when the user provides a link and wants the original content, evidence, transcript, summary, or source-grounded analysis. |