headlines

name: headlines category: information description: Fetch top news headlines from major international RSS feeds (Reuters, AP, BBC, etc.). No API key required. Tools: http. test_prompt: What are today's top headlines? permissions: - android.permission.INTERNET

Headlines Skill

Fetches top news headlines from public RSS feeds of major international news outlets and press agencies. No API key required – RSS is free and public.

Tools

RSS Feed Sources

🌍 International / Global

🇩🇪 Germany (Deutsch)

🇦🇹 Austria (Österreich)

🇨🇭 Switzerland (Schweiz)

🇬🇧 United Kingdom

🇫🇷 France

🇺🇸 USA

🇮🇹 Italy / 🇪🇸 Spain

Topic-specific BBC RSS feeds

How to fetch headlines

Basic RSS fetch

{
  "method": "GET",
  "url": "https://feeds.bbci.co.uk/news/rss.xml",
  "response_format": "text"
}

The response is XML. Key parsing rules:

• Each news item is inside <item>...</item> (RSS) or <entry>...</entry> (Atom, e.g. Heise)
• Extract: <title>, <description>, <pubDate> (or <updated> for Atom), <link>
• Ignore the first <title> tag – it belongs to the <channel> (feed name), not a news item
• Strip HTML tags and <![CDATA[...]]> wrappers from descriptions
• Present max 5–10 headlines to avoid overwhelming the user

Workflows

"What are today's top headlines?"

1. http → GET https://feeds.bbci.co.uk/news/rss.xml with response_format: "text"
2. Parse XML: extract first 5–7 <item> titles and descriptions
3. Summarise: "Here are today's top headlines from BBC News: 1. … 2. … 3. …"

"Read me the German news"

1. http → GET https://www.tagesschau.de/xml/rss2 with response_format: "text"
2. Parse first 5 items
3. Summarise in the user's configured language

"What's happening in Austria?"

1. http → GET https://news.orf.at/stories/rss/ with response_format: "text"
2. Parse and summarise top 5 stories (ORF publishes large amounts of APA wire content)

"Tech news"

1. http → GET https://www.heise.de/rss/heise-top-atom.xml with response_format: "text"
2. Note: Heise uses Atom format – items are in <entry> tags, not <item>
3. Summarise top 5 tech stories

"Give me a global news overview"

Fetch 2–3 feeds from different regions, deduplicate overlapping stories, present a combined summary:

1. http → GET BBC RSS (global English perspective): https://feeds.bbci.co.uk/news/rss.xml
2. http → GET AP RSS (US/wire perspective): https://apnews.com/hub/ap-top-news?format=rss
3. http → GET Al Jazeera RSS (alternative perspective): https://www.aljazeera.com/xml/rss/all.xml
4. Merge, remove duplicates, summarise 7–10 unique headlines

"Read me the news every morning at 7"

1. datetime → absolute with base: "tomorrow", time: "07:00", output_unit: "milliseconds"
2. scheduler → create with:
   • label: "Daily News Briefing"
   • instruction: "Fetch the top 5 headlines from https://feeds.bbci.co.uk/news/rss.xml using the http tool (method: GET, response_format: text). Parse the XML and extract the titles. Then read them aloud via TTS."
   • trigger_at_ms: timestamp from step 1
   • recurrence_type: "daily"
   • recurrence_time: "07:00"
3. Confirm: "I'll read you the BBC headlines every morning at 7."

"French news in French"

1. http → GET https://www.france24.com/fr/rss with response_format: "text"
2. Parse and summarise in French (AFP wire content via France 24)

"News from multiple countries"

Fetch one feed per requested country:

• Germany → Tagesschau
• Austria → ORF
• Switzerland → SRF
• Global → BBC or AP

Recommended source combinations

Deutsche Welle language codes: en, de, fr, es, ar, pt, ru, zh, uk, tr, id, sw, ha, am, bn, fa

Output format rules

• Voice (TTS): Read 3–5 headlines. One sentence each. Name the source: "According to BBC News: …"
• Text response: List 5–10 headlines with source name and one-line description
• Always mention the time/date of the headline if available from <pubDate>
• Never dump raw XML at the user – always summarise in natural language
• If a feed returns an HTTP error, fall back to an alternative source
• Summarise in the user's configured language, even if the feed is in a different language
• Atom feeds (e.g. Heise) use <entry> instead of <item> and <updated> instead of <pubDate>

Notes

• Always use response_format: "text" for RSS feeds (they return XML, not JSON)
• No auth_provider needed – all listed feeds are public
• The http tool truncates responses at 5000 characters – sufficient for 10–15 headlines
• For best voice results, fetch only 1 feed and read 3–5 headlines
• Fetch multiple feeds only when the user asks for a cross-source overview