mirror of
https://github.com/Druthulu/BFM-decomp
synced 2026-09-30 23:37:39 -04:00
Drew T
443279e25a
- Fleet 95.42% fn / 92.4% instr / 85.5% distinct; phase arc +3.42 / +4.9 / +7.5pp; R22 140/140 x9. - ~3,500 members in 11 commits, from TWO engines that both paid: deterministic tooling repair (same population scored 842 pre-fix vs 1,582 post-fix; cookbook §134) and two-lane agent waves (bank rate 83->93->89%, reconcile lane 12/12; cookbook §135). - Queue RE-DERIVED at HEAD after the final bank (not carried forward): the >=20-member/<=60-ins B-shape is EXHAUSTED; 33 families/34,895 ins remain at 61-120 ins and 111/33,554 at 10-19 members; S6c's 9 jr families (9,364 ins) still untouched. Unmatched fleet-wide 15,161 instances/935,927 ins. - Flags for a fresh session: (1) the T6 ROI-floor trigger is REFUTED — do not close P30 on it; (2) measured token economics — deterministic ~0 tokens/+0.5pp vs wave-3 4.1M/+0.3pp; probe the cheap lever before funding the expensive one.
phase-ends/ — the living record
PhaseEnd_Phase[N].md— written at each phase boundary (format defined inPROJECT_CONTEXT.md). Append-only history: build log, deviations, commit message, rules added, changelog. Never deleted.CURRENT_PHASE.md— the in-phase autonomous log: approved phase plan, per-task checkboxes, current task pointer, blockers. Created at phase start, updated after every task, absorbed into the PhaseEnd file and deleted at phase close. This is the crash/compaction recovery point.- Sessions read
PROJECT_CONTEXT.md, then everyPhaseEnd_*.mdin numeric order, thenCURRENT_PHASE.md(if present) — in that order, every session.