2 Commits

Author	SHA1	Message	Date
Drew T	5f07d099e6	docs(phase-26): finalize Task-6 triage (119 agents) + banking caveat; pause before Task 8 ... - full triage complete: cheap 29 (23 closeness-0 isolation-MATCH) / permuter 29 / fable5 61 (1.71M ins). - attempted to bank the 23 cracked wins into ov077 -> 0/23: the match_one isolation-MATCHes are genuine function matches but carry standalone struct/scalar typedefs + Ghidra-typed sigs that conflict with the real ov077 TU (redefinition of struct Obj / conflicting types) = the §41 def-side wall. Banking needs the Task-8 --reconcile / canon_sig_reconcile pass (not run — paused before Task 8 per Drew). - docs/phase26-triage.md carries the crack curriculum + the caveat; seeds in .run/phase26-seeds/. - src pristine, ov077 byte-identical.	2026-07-11 23:12:53 -06:00
Drew T	80536efbea	feat(phase-26): task 6 — Ultracode triage of 119 draftable substantial families ... - .run/wf_triage*.js Workflow: per-family m2c draft (+§8 jtbl handling) -> match_one reloc-masked closeness -> classify cheap/permuter/fable5 + §31/§45 lever + seed to .run/phase26-seeds/. - 119 families triaged (2.38M templatable ins): cheap 22 (324k ins, 20 already isolation-MATCH) / permuter 23 (340k) / fable5 74 (1.72M). The cheap-Opus triage flywheel cracked 20 families to closeness-0 as a side effect (each templates x134) -> Task-8 gate+template candidates. - docs/phase26-triage.md = the crack curriculum (already-cracked seeds + top-30 Fable5 targets + permuter band). Seeds in .run/phase26-seeds/. match_one closeness is an INDICATOR; the whole-binary byte-gate (Task 8) is the sole arbiter (G3/P9).	2026-07-11 23:00:26 -06:00