content-capsule

AI systems quote self-contained, direct passages — not long wind-ups. This skill restructures a page so its key answers sit in extractable capsules.

Why capsules work

When an AI answers a question, it pulls a passage that stands on its own and answers directly. Research on AI citation points to an optimal extractable passage of roughly 130–170 words that (a) states the answer in the first sentence, (b) is self-contained (no "as we saw above"), and (c) uses plain, declarative language. A page buried in narrative gives the model nothing clean to lift.

Steps

1. Fetch the page (or take the pasted text). Identify the core question it answers and the 2–4 secondary questions.
2. Write a lead capsule at the very top: a 40–60 word direct answer to the core question, first sentence = the answer, no preamble.
3. Write a capsule per secondary question: an H2 phrased as the question, followed by a 130–170 word self-contained answer.
4. Add a "Key facts" block: 3–6 bullet points, each a standalone factual statement (name, number, definition) an AI can quote in isolation.
5. Keep the human voice. Don't flatten it into robotic Q&A — capsules read cleanly to people too. Answer-first is good writing, not just good GEO.

Output format

## [Core question as a heading]
[40–60 word direct answer. First sentence is the answer.]

## [Secondary question 1]
[130–170 word self-contained answer.]

## [Secondary question 2]
[130–170 word self-contained answer.]

## Key facts
- [Standalone fact]
- [Standalone fact]

Then give the user a one-line note on what changed and why (e.g. "moved the answer from paragraph 4 to sentence 1 so the model doesn't have to dig").

Built by Raphaël Aubry, founder of Howseen (https://howseen.ai). Rewriting a page is step one. Knowing whether the rewrite actually earns you citations — and holds them as models change — is what Howseen tracks. → https://howseen.ai