@pipeworx/bing-news

Connect: https://pipeworx.io/mcp — every tool in the catalog, including @pipeworx/bing-news’s. Install: one-click buttons

No MCP client? Skip the connection: POST https://gateway.pipeworx.io/v1/tools/search_packs {"query":"..."} to find a tool below, GET /v1/tools/<name> for its schema, POST the same URL with arguments for the data — see For AI agents.

Tools: 1

Keyword news search via Bing News RSS — a third, independent news source (different vendor and infrastructure from Google News/country-news and GDELT/gdelt), returning recent matching articles ranked by Bing with the real article link, publish time, source publication and a short description.

Tools

• bing_news_search(query, count?) — searches Bing News for query and returns up to count (1-50, default 20; limit, num and max_results are accepted aliases) matching articles: title, link (the real destination article URL, not Bing’s tracking redirect), description, published (RFC-822), source (the publication name). count is an exact upper bound. The response always carries requested, returned and exhausted; when Bing has fewer matching articles than asked for, exhausted is true and note says how many exist.

Auth

Keyless. No registration, no API key, no rate-limit documented by Bing for this endpoint.

How the count is met

Bing’s default ranking returns very short pages for narrow or generic queries (as few as 1). When the first page is short the pack widens through Bing’s date-sorted and recent-window views of the same query, de-duplicating by article URL and keeping only articles whose title or description contains a query word, so the count is never padded with unrelated stories. Bing’s own ranked results always come first.

Data sources

• https://www.bing.com/news/search?q=<query>&format=rss — Bing’s own news search, RSS output. Not a documented/versioned API — it is the same feed a browser gets from the “News” tab’s RSS link, so its shape can change without notice the way any scraped HTML/XML can.

Rediscoverable-the-hard-way notes:

• The link element is a tracking-redirect wrapper, not the article URL. Bing wraps every result as https://www.bing.com/news/apiclick.aspx?ref= ...&url= followed by the percent-encoded real article URL and more tracking params. The real destination is the value of that url query param, and it is present and decodable on every item observed (verified live 2026-10-09 against 5 search results — msn.com, finance.yahoo.com, pulse2.com). realArticleUrl() decodes it; this pack returns the decoded URL as link, never the apiclick wrapper. If Bing ever omits the url param, the code falls back to the raw link rather than dropping the item.
• The source element is namespaced. It is News:Source, not a plain source element — easy to miss if you copy a generic RSS-item parser from another pack (country-news’s Google News feed has no equivalent element at all; it comes from a different field).
• A 200 is not proof of a real result. Like GDELT and Google News, a CAPTCHA/anti-bot challenge page can come back with HTTP 200 and content-type: application/xml or text/html instead of the RSS feed — guarded here by checking for a literal <item tag before parsing, so that case surfaces as a loud failure rather than a silent empty result.
• Cloudflare Workers egress is NOT blocked for this endpoint — edge-probed live 2026-10-09 from a throwaway wrangler dev --remote Worker on the prod account: bing.com/news/search answered the Worker’s own egress HTTP 200, 10,393 bytes, well-formed RSS. Unlike country-news (Google News blocks CF egress outright), this pack’s direct fetch is not currently routed through the shared Supabase egress relay — the relay-fallback code exists (same _proxyUrl/_proxyToken convention as country-news/gdelt) but nothing in the gateway injects those args for this pack’s slug, because nothing has needed to yet. Re-probe before assuming that’s still true months later.

Tools

• bing_news_search — Keyword news search via Bing News, keyless.