Skip to content

Memory

The Memory tab lists the brain’s own memory corpus — a shared corpus of execution-verified audit findings (patterns, not code) that every corral certify --local run can draw on. An entry doesn’t get in by a model asserting it; it gets in because a run actually caught it — a mutant a suite missed, a dead assertion a critic flagged and a jail confirmed was really inert. See memory tiers and the learning loop for how an entry moves from advisory to vetted. Every entry links “▸ see it in DuckDB” back to the run that proved it.

The Memory tab listing three advisory memory entries: “a length-only password test misses the digit/uppercase rules” (advisory: test coverage, linked to the passwd.go run), “range==tuple is always False in Python 3 — a dead assertion” (advisory: dead-test pattern, linked to the full_permutation run), and “a 100%-passing suite still missed 40% of planted faults — kill-rate, not pass-count, is adequacy” (advisory: adequacy framing), each tagged CORRALAI · FINDING with a “▸ see it in DuckDB” link, plus a search box and a tier filter above the list