How to curate a newsletter with a news API#

To curate a newsletter, run one saved APITube query per issue: scope it to your topic, limit it to the days since the last issue with published_at.start, and gate quality with source.rank.opr.min. The response is the issue’s candidate list — every article with its title, link, summary and source — ready for a human pick or an automated template. The same query, re-run weekly, is the entire pipeline.

curl -G "https://api.apitube.io/v1/news/everything" \
  --data-urlencode "topic.id=industry.crypto_news" \
  --data-urlencode "published_at.start=NOW-7DAYS" \
  --data-urlencode "source.rank.opr.min=6" \
  --data-urlencode "language.code=en" \
  --data-urlencode "per_page=25" \
  --data-urlencode "api_key=YOUR_API_KEY"

NOW-7DAYS is the relative date syntax, so the window slides automatically — no date arithmetic in your scheduler.

How do you scope the query to your newsletter’s topic?#

Pick the narrowest filter that matches how your topic is defined:

• A subject area — topic.id (like industry.crypto_news above) or an IPTC category via category.id; see how to filter by topic.
• A keyword in the headline — title=electric vehicles; multi-word values need every word to appear.
• A company or person beat — organization.name / person.name matches detected entities rather than raw text, so a “Nvidia this week” section doesn’t miss articles that say “the chipmaker” in the headline.

Filters combine with AND, so topic.id + title gives you “this subject, but only stories about X” — a natural fit for newsletter sections. Run one query per section rather than one broad query you re-sort by hand.

How do you keep the issue high-quality?#

Reader trust dies on link rot and content farms, so push quality into the query instead of eyeballing every candidate:

• source.rank.opr.min=6 sets an authority floor (OPR is 0–10; raise it for a stricter list) — or use the stricter composite flag described in filtering to trusted, high-quality sources.
• ignore.source.domain=example-tabloid.com drops up to three publishers your readers dislike, comma-separated.
• Wire reprints cluster around one event; fetch the story cluster for a headline you picked and feature the strongest version instead of three copies of the same news.

How do you hand the result to your newsletter tool?#

Three integration shapes, cheapest first:

• RSS: append export=rss and paste the URL into any tool that ingests feeds (most email platforms do) — the API returns the same search as an RSS 2.0 document. See how to export search results as RSS.
• JSON: call the API from your build script and render candidates into your template — each article carries title, href, description, published_at and source, which is everything an issue block needs.
• Spreadsheet triage: the response’s export block includes ready CSV/XLSX URLs, so an editor can pick winners in a sheet before layout.

The full list of filters you can bake into the saved query is in the official News API parameter reference.

Common Questions#

• How do I avoid repeating an article across two issues?
• How many requests does a weekly newsletter cost?
• Can I get article summaries for the issue blurbs?

How do I avoid repeating an article across two issues?#

Anchor the window to your publishing cadence: a weekly issue queries published_at.start=NOW-7DAYS at the same hour each week, so consecutive windows touch without overlapping. If your schedule slips, store the published_at of the newest article you featured and pass it as the exact published_at.start of the next run.

How many requests does a weekly newsletter cost?#

One search that returns articles counts as one request against your quota, so a five-section weekly newsletter costs about five requests per issue — around twenty per month, which fits comfortably in any plan. Requests that return zero articles are not charged. See what counts as a request.

Can I get article summaries for the issue blurbs?#

Yes — each article object already includes a summary produced during processing, alongside description. Render it as the blurb draft and edit by hand where it matters; for tighter payloads, request only the fields you use with the fl parameter — see how to select specific response fields.