Use when you have a week of saved links and need an issue with a view rather than a list.
curated-newsletter-issue-build.md
You are assembling a curated newsletter issue. The value is the judgement between the links, not the links.
Saved items, each with title, source and my note: {{SAVED_ITEMS}}
Who subscribes and what they do for a living: {{SUBSCRIBERS}}
Issue length and how many items it holds: {{ISSUE_SHAPE}}
The through-line I think this week has, if any: {{THEME_HUNCH}}
Output:
1. Selection table: Item from {{SAVED_ITEMS}} | Why a reader in {{SUBSCRIBERS}} would care | Overlaps with another item | In, out or reserve.
2. Running order for the items marked in, up to the count in {{ISSUE_SHAPE}}, with the reasoning for the first slot and the last.
3. The issue itself: an opening under 120 words that makes an argument, then each item with a two to three sentence take that says the thing my note implies but does not spell out.
4. Verdict on {{THEME_HUNCH}}: supported, partly supported, or not at all. If not, say what this week is actually about.
Rules:
- Do not summarise an item beyond what {{SAVED_ITEMS}} contains. You have not read the sources.
- Cut rather than pad to reach the count in {{ISSUE_SHAPE}}.
- Nothing gets in on novelty alone. It has to be useful to {{SUBSCRIBERS}}.
- No em dashes. British spelling.
Replace each placeholder with your own detail. The more specific you are, the less the model invents.
When should I use this rather than summarising each link?
When the issue is meant to have a view. Summaries of each link are the version subscribers skim, because the judgement between items is what they subscribed for. The prompt is told it has not read the sources, only your notes, so it makes selection calls rather than pretending to review.
What makes the takes worth reading?
The saved items with title, source and your own note on each, who subscribes and what they do for a living, the issue length and item count, and the through-line you think the week has. Write the note when you save the link. The takes are only as good as those notes.
What does it assemble?
A selection table marking each item in, out or reserve with overlaps flagged, a running order reasoned for the first and last slots, the issue itself with an opening under 120 words that argues something, and a verdict on your theme hunch. Keep the reserve items for a thin week.
What makes an issue read badly?
Forcing the theme when section 4 says it is not there. An honest grab bag reads better than a manufactured through-line and readers can see the seams. Padding to reach the item count is the same error in another form, and the prompt is told to cut instead, so let it.