news-briefing

name	news-briefing
description	Fetch and summarize current news/headlines for any country, language, or topic using Google News RSS via curl. Use when the user asks for latest news, today's headlines, a news digest, or topic-specific news summaries.

News Briefing via Google News RSS

Class-level workflow for building current-events summaries from live RSS feeds.

Why this approach

• Individual outlet RSS URLs are unreliable — guessed paths like detik.com/rss or kompas.com/getrss/nasional return 404. Go to the aggregator first.
• The mcp__fetch__fetch tool enforces robots.txt, and news.google.com disallows it. Use curl in the terminal instead; it succeeds where fetch refuses.
• After ~3 consecutive fetch-tool failures the MCP server enters an auto-retry cooldown (~60s). Don't hammer it — switch to curl immediately.

Workflow

1. Build the locale feed URL:
   • Headlines: https://news.google.com/rss?hl=<lang>&gl=<country>&ceid=<country>:<lang>
     • Indonesia: ?hl=id&gl=ID&ceid=ID:id
     • US English: ?hl=en-US&gl=US&ceid=US:en
   • Topic/search: https://news.google.com/rss/search?q=<urlencoded query>&hl=..&gl=..&ceid=..
2. Fetch with curl and a browser UA, save to /tmp: curl -sL --max-time 25 -A "Mozilla/5.0" "<URL>" -o /tmp/feed.xml
3. Parse with Python stdlib regex — use scripts/gnews_parse.py from this skill (handles unescaping, <source> extraction, trailing-outlet dedup): python3 scripts/gnews_parse.py /tmp/feed.xml 25
4. Summarize grouped by section (e.g. Nasional / Ekonomi / Internasional / Hiburan), keeping headlines close to original wording.
5. Reply in the user's language.

Output rules (user-checked)

• Attribute the source media for every headline — this user asks "source nya darimana?" when provenance isn't obvious. Tag each item with its outlet and list the feed URL at the end of the reply.
• Note the retrieval date/time so staleness is clear.
• Strip the trailing " - Outlet" duplicate from titles when the <source> tag already names the outlet.
• Filter junk: SEO/spam entries (gambling etc.) do slip into aggregated feeds — drop them silently.

Pitfalls

• Titles in Google News RSS are HTML-escaped — run them through html.unescape.
• <source> carries attributes (<source url="...">Outlet</source>) — capture inner text, not the raw tag.
• If curl also fails, fall back to fetching 2–3 major national outlets' homepages directly rather than retrying dead RSS URL guesses.