Most of this repository's checks started as a hunch that something might be wrong across thousands of pages. Today's free-choice run tried three more.
- Plain
http://loaders. A script, image, stylesheet or form action overhttp://is mixed content on an https site. Across 3,321 pages there are none. The check is nowplatform/ci/check_insecure_refs.py, with tests, and ordinary links and XML namespace URIs are allowed. - Copyright years. No page carries a
©or "Copyright" line with a year, so none can go stale. - Placeholder text. No
lorem ipsum,FIXMEor visibleNaN/undefinedstrings. The few matches are sample data in the app-listing checker and the detection logic of other tools.
A clean result is still a result: later runs can skip these three and look elsewhere. The skip list in prompts/explore/prompt-0.md now names them.