bbc-news

This is one script with an optional section argument, reading BBC feeds for UK, world, technology, business and other sections. It needs no authentication and no API key, which makes it a good default for headline retrieval inside a briefing job. The trade-off is inherent to RSS: there is no control over story ordering or fields beyond what the feed carries, and freshness is exactly the feed's freshness, so a prompt asking for recent coverage can be satisfied by items that are hours old. Section names must match the script's expected slugs rather than display names.

Use case: Pulling BBC headlines by section into a news digest or morning briefing.

# Sections are the whole interface, including its limits This skill fetches stories from news feeds by section: top, world, uk, business, politics, health, education, science, technology and entertainment, plus UK regional and world-region variants. The interface is thin, and almost all of the practical difficulty comes from how feeds behave rather than from how you call the script. ```bash python3 scripts/bbc_news.py technology python3 scripts/bbc_news.py --list ``` ## Enumerate before you hard-code Feed section names change. Call the list option and cache the result before writing a job that depends on a specific section name, otherwise a rename turns into a silent empty digest. ## Deduplicate across sections Top stories are aggregated, so a technology story can appear in both the top section and the technology section. If your job pulls two sections and concatenates them, expect overlap and dedupe on the story link before summarising. ## Handle staleness explicitly A feed that has not changed since the last poll returns the same items. A scheduled digest built on it will happily send you yesterday's headlines as today's. Record the identifiers you have already reported and diff against them, rather than trusting the fetch. ## Keep the fetch thin Fetch, dedupe, summarise. Resist enrichment inside the same job; the moment a section fetch also tries to resolve article bodies, timeouts on one article start failing the whole digest.