Skills on AI

Reading Capture

Download raw SKILL.md

Use when someone reads books, articles, or papers and forgets most of it, or wants a system to capture highlights and ideas and then process them into their own words so they can be found and used later. Distinct from [[ai-note-system]] (managing notes with an AI assistant), [[deliberate-practice]] (building a skill through focused repetition), and [[tab-management]] (holding pages to read).

Highlighting feels like progress, but a pile of highlights is only a pile. Reading capture is a two-stage habit: collect what caught your attention while reading, then later process it into your own words, filed where you will look for it. This skill sets up both stages for the user's kind of reading.

Input required

  • What they read: books, articles, papers, newsletters, e-books, paper.
  • Why they read: pleasure, work, study, a project, or a decision.
  • Where capture happens: a reading app, a notes app, paper, margins.
  • How notes are used later: writing, teaching, decisions, memory.
  • Time available for processing each week.

Questions to ask first

Ask in one batch of at most five questions, skipping any already answered:

  1. What do you read most, and in what form (paper, e-reader, web)?
  2. What do you want to be able to do with it afterward?
  3. Where do your highlights and notes go now, and do you ever revisit them?
  4. How much time can you give processing per week?
  5. Do you have a notes app or folder you already trust?

Method

  1. Decide what to capture. Before reading, write a question or purpose in a line. Capture what answers it, changes your mind, surprises you, or you may need to quote. Skip the rest.
  2. Capture lightly while reading. Highlight or mark the passage, and add a three-word reason if possible ("example of X", "disagree", "use for project"). Note the page or location. Keep reading flowing; do not stop for full notes.
  3. Keep one landing place. Highlights from every source arrive in one inbox, with title, author, and date, so nothing is stranded in a device or margin.
  4. Process within a few days. Reread the highlights and ask of each: keep, cut, or turn into a note. Aim to cut most. Memory of the context fades, so processing soon after is easier.
  5. Rewrite in your own words. For each kept item, write a short note: what the idea is, in a sentence or two, without copying. Add why it matters to you and one link to something you already know or a project. If you cannot restate it, you have not understood it; mark it to reread.
  6. Mark quotes properly. If you keep an exact quote, put it in quotation marks with the source and location so it is not mistaken for your own words later.
  7. File for retrieval. Give each note a title that states the idea, a few consistent tags or a project link, and a link to the source. Ask: what would I search for when I need this?
  8. Use it. Review notes when starting a project, writing, or deciding. Notes that never get used suggest capturing less or tagging to match real uses.
  9. Close the book. Finish each book or paper with a short summary of its main points and what you will do differently, if anything. For the habit of applying what you read, see Deliberate Practice.

Output format

  • Capture rules: what to mark and how, in a few lines.
  • Processing routine: when, how long, and the keep/cut/note test.
  • Note template: title, own-words summary, why it matters, link, source, tags.
  • Filing scheme with a small tag set.
  • Weekly review slot.

Ground rules

  • Own words, honestly. Distinguish quotes from paraphrase; do not present another author's words as the user's.
  • Do not invent source details. Quotes, pages, and attributions come from the actual text; if unsure, mark it to check.
  • Respect copyright and licenses. Short quotes with citation for personal notes are normal; do not copy whole chapters or share protected text freely.
  • Capture less. A short list processed beats a long list ignored.
  • Tools are examples only.

Self-check before answering

  • Is there a stated purpose for the reading?
  • Is there a single capture inbox and a processing date?
  • Does the process force restating in the user's own words?
  • Are quotes labeled and sourced?
  • Will the user realistically find the note again?

Common pitfalls

  • Highlighting everything — nothing stands out.
  • Capture without processing — a graveyard of highlights.
  • Copying instead of rewriting — no understanding gained.
  • Notes titled by source only — "Chapter 3" is useless in search.
  • Too many tags — retrieval becomes guesswork.
  • Over-processing light reading — save the effort for what matters.

Example prompt

I read a few books a month and lots of articles, and I forget almost everything. Set up a capture and processing routine: what I highlight while reading, how I turn it into my own words, and how I file it so I can find it later.

Learn more

View reading-capture/SKILL.md on GitHub