naver-news-search

Queries the Naver Search Open API news endpoint through a hosted proxy (k-skill-proxy) to conservatively assemble the latest Korean-language news for a search term, returning article title, publish time, link, and summary. A key convenience is that users do not need to obtain their own Naver developer credentials: the upstream Client ID and Secret are injected by the proxy server, so callers simply hit the proxy's public, read-only endpoint.

For each result it returns the article title and summary description, both with Naver's bold-highlight tags and HTML entities already stripped by the proxy, the original RFC822 publish time plus a parsed ISO-8601 UTC timestamp, the Naver news redirect link, and the original source link. Query parameters include the search term of at least two characters, a display count from 1 to 100 defaulting to 10, a one-indexed start position, and a sort of either relevance, which is the default, or date for newest-first; the start and display combination cannot exceed Naver's 1000-item search window. A typical workflow confirms the search term, chooses the sort based on whether the user wants latest results, calls the endpoint, and summarizes the top three to five items with title, publish time reformatted to Korean local time, summary, and link.

It is meant for finding and summarizing recent news by keyword, and explicitly not for paywalled or login-only articles, full article bodies since only the summary description is available, real-time stock, currency, or real-estate quotes, or any path requiring block or CAPTCHA circumvention. Documented failure modes cover bad requests, an unconfigured or misconfigured proxy, exceeding the daily quota of 25,000 calls, and upstream errors, with a firm no-retry policy on blocks or outages. Privacy rules keep it from permanently storing queries or results, requesting article bodies, or adding speculation beyond the article text itself.