A Codex skill for strict high-signal content extraction.
It turns articles, X/Twitter threads, Xiaohongshu posts, screenshots, and OCR text into concise Traditional Chinese knowledge cards. The skill is intentionally selective: it preserves actionable, insightful, reusable ideas and rejects fluent but low-value content.
- Extracts high-quality ideas from long-form and social content
- Scores content by actionability, insight, and reusability
- Produces one searchable Markdown knowledge card by default
- Keeps source evidence in the original language
- Handles screenshots with strict visible-text rules
- Outputs a short rejection when content is not worth saving
Use this skill when you want Codex to:
- Extract high-signal ideas from an article or thread
- Judge whether a piece of content is worth saving
- Turn a screenshot into a knowledge card
- Convert Xiaohongshu or X/Twitter content into reusable notes
- Filter low-value content before it enters your knowledge base
Example prompts:
Use $high-signal-extractor to turn this thread into a knowledge card.
這篇值得保存嗎?幫我用 high-signal-extractor 萃取高信號內容。
幫我看這張截圖有沒有值得保存的洞察。
The skill evaluates each candidate idea on three criteria:
可行動: leads to a concrete next step, decision, experiment, question, or operation洞察性: reveals causality, a counterintuitive point, a tradeoff, or a hidden pattern可複用: can become a principle, checklist, framework, template, or decision rule
Content normally needs at least two criteria scoring 4/5 or higher to become a saved card.
For strong content, the skill produces:
- A concrete, searchable title
- Save judgment:
必存or可選保存 - Quality scores
- One-sentence insight
- Source evidence and context
- Reusable principle
- Applicable and non-applicable scenarios
- Suggested future uses
- Tags
For weak content, it produces a short 不建議保存 rejection with reasons and what would need to be added for the content to become worth saving.
Clone this repository into your Codex skills directory:
mkdir -p ~/.codex/skills
git clone https://github.com/YSKMsky/high-signal-extractor.git ~/.codex/skills/high-signal-extractorThen start a new Codex session so the skill metadata can be discovered.
SKILL.md: the skill definition and workflowagents/openai.yaml: UI metadata for Codex