[srinivasgumdelli]

An audit of every closed discovery issue (Aug to Nov 2026) found false rejections with a common cause: a shortcut, not a judgement. This adds five checks to run before any out of scope outcome, each tied to the case that taught it:

1. Open every roundup. The Discovery candidates - 2026-09-22 #71 recap hid Plugin4Shell.
2. Judge against existing records, not the article's framing. Discovery candidates - 2026-10-02 #88 was rejected against the 0075 precedent; those episodes are now 0114-0116.
3. Sweep every target a primary names. Transluce's Data USA and UNM episodes went untriaged.
4. Weigh a dispute; don't count it. Case: 0113, judged against the 0038/0051 standard.
5. Re-run a changed scope rule on existing records. Case: the 0061 and 0084 withdrawals.

Documentation only.

🤖 Generated with Claude Code