We need to produce valid JSON with fields: seo_title, summary, key_entities, expert_faq, internal_link_anchor, primary_keyword, secondary_keywords, content_type.

We must write a polished summary 180-250 words, 3-4 paragraphs separated by \n\n. Each paragraph 45-65 words. Must bold 3-5 key details across entire summary using double asterisks. Must not bold generic words. Must not use banned vocabulary etc.

We must include key_entities section with companies, people, locations only mentioned in source. Source mentions "unix.foo"? Actually headline is "The Last People Who Know How It Works · unix.foo". Source is Hacker News. No specific company names? It mentions "modem", "AI assistant", "kernel", but not specific company names. It mentions "graybeards"? Not a company. It mentions "the last people". No explicit company names. Maybe "unix.foo" is a domain, but it's not a company. The source mentions "Cyrus Lopez"? Actually author is "BYCYRUS LOPEZ". So a person: Cyrus Lopez. Also maybe "unix.foo" is a location? It's a domain. Could be considered location? Probably not. But we can include "unix.foo" as location? The instruction: key_entities: companies, people, locations only mentioned in source. So we can list "people": "Cyrus Lopez"; companies: ; locations: "unix.foo"? It's a domain, maybe considered location? Might be okay. But better to include "unix.foo" as location? It's not a location per se. Could be "Hacker News" as location? Not mentioned as location. The source is Hacker News, but not necessarily a location. Could include "Hacker News" as organization? It's a site. Might be considered a company? It's a platform. But it's not mentioned as a company; it's the source. The instruction says only entities explicitly mentioned in source. The source text includes "unix.foo" and "Cyrus LOPEZ". Also "Hacker News" is the platform but not part of the article content? The headline includes "· unix.foo". That is part of the headline, so it's mentioned. So we can list