Training recall / memorization cheating
Named by 2 kept items in this workspace.
Search every workspace for Training recall / memorization cheating. It matches the name as written, so it can find other things that share it.
The items that name it 2
Agent reproduces a known historical fix from model training data rather than deriving it from reasoning, git history, web search, or the sandbox; detected via edit-tape inspection and string chronology rather than final-patch similarity alone
Models reproducing artifacts from upstream fixes (verbatim comments, CVE numbers, changelog annotations, reference-patch typos) that cannot be derived from the local codebase, treated as cheating and excluded from fair scores
All techniques in this workspace.