v1.0.0 — **PHASE 3 COMPLETE · 1.0.0 SHIPS**

Added — T-308 post-clear optimization mode

• analysis/optimization.py::post_clear_targets_for_encounter() synthesizes the three optimization levers — burst alignment (T-105), GCD drops/min (T-008), raid DPS vs target (T-204) — into a composite polish_score (geometric mean of each, 0..1). Per-kill leaderboard sorted worst-first surfaces the biggest gap to fix next.
• Watchlist-scoped — kills not in our watched reports are excluded (don't optimize against random ranking-board pulls).
• Score curves: burst saturates at ≥85% (industry-standard "great" alignment), GCD drops/min from 0 (perfect) → 6 (Pict-tier movement-heavy floor), DPS ratio 0.8 (floor) → 1.0 (matches median).
• API: GET /api/encounters/{id}/post-clear-targets.
• 7 new tests (4 pure-function score curves + composite-with-Nones + no-kills-note + live FRU structure). 281 tests total.

1.0.0 ships. All Phase-3 tasks (T-301..T-308) shipped + stable. Per CLAUDE.md §4, the full planned feature set is complete:

• Phase 0: foundation + Mode-1 dashboard (T-001..T-011) — ingest, schema, M-WIPE, M-FAULT-Mode-1, M-GCD, dashboard
• Phase 1: capture + early inference (T-101 + T-103..T-108) — live polling, phase segmentation, consensus, M-BURST, M-PARSE, member resolution, ability classifier
• Phase 2: cross-group field comparison (T-201..T-208) — field backfill, fight_model persistence, mechanic classification, empirical DPS check, prog-vs-field, M-CART, M-GATE, Encounters UI
• Phase 3: strat-aware + polish (T-301..T-308) — strat_config editor, M-FAULT Mode-2, M-MIT, fault disambiguation, M-RECOV, M-CONS, M-REPORT, optimization mode

Phase 1 backlog still open (not gating)

• T-109 Tighten T-004 combatant filter for Ultimate reports (scheduled; downstream workarounds handle it).
• T-309 Drag-and-drop visual strat editor (polish on T-301; defer until needed).