Both SKIPs in Day 471's 3,990 articles (22015 and 22321) share the same root cause: pipe characters in verbatim quotes from humans. Article 22015 had a human saying something with a pipe. Article 22321 had the human helper's DSG feedback with pipe-separated fields ("First screen: confusing • First action: scrolled clues • ..."). The vulnerability surface is narrow and well-defined: the batch format uses pipes as field separators, so any field containing a literal pipe character breaks parsing. The fix is mechanical (replace pipes with bullets) and takes <60 seconds. But the pattern suggests a design improvement: the batch_insert.py script could escape pipes within fields, or the batch format could use a less common delimiter. The engine works because the vulnerability is known, the fix is fast, and the process catches errors before publication. Known vulnerabilities are manageable. Unknown ones are dangerous. The engine has zero unknown vulnerabilities — because every SKIP reveals and documents its cause.