diff --git a/.agent-plan.md b/.agent-plan.md
index de08608..e6210b1 100644
--- a/.agent-plan.md
+++ b/.agent-plan.md
@@ -6,7 +6,7 @@
 
 ## Current System State
 
-**v0.5.0 in progress — Milestones 7–11 complete, v4-M1 in PR.** Full simulation engine + render/bundle + exposure filtering + CLI commands + validation harness implemented. v4 engine changes + build pipeline ready. 609 tests passing.
+**v0.5.0 in progress — Milestones 7–11 complete, v4 dataset shipped.** Full simulation engine + render/bundle + exposure filtering + CLI commands + validation harness implemented. v4 engine changes + build pipeline merged (PR #21). v4 dataset generated and validated. 609 tests passing.
 
 ---
 
@@ -24,7 +24,7 @@ See `docs/v4/design.md` for full details.
 - [x] `scripts/spike_category_signal.py` — spike experiment validating category signal approach
 - [x] Updated `CLAUDE.md`, `AGENTS.md`, `.agent-plan.md`
 
-### v4-M1: Engine + build pipeline ✓ (in PR)
+### v4-M1: Engine + build pipeline ✓ (PR #21, merged)
 
 Engine changes:
 - [x] Add `category_latent_correlations` to `difficulty_profiles.yaml` (intro profile, scale 1.8)
@@ -37,17 +37,19 @@ Engine changes:
 - [x] 19 new tests (16 windowed snapshot + 3 category-latent correlations)
 
 Build pipeline:
-- [x] `scripts/build_v4_snapshot.py` — day-14 snapshot + leakage trap + structured missingness + subsampling
+- [x] `scripts/build_v4_snapshot.py` — day-10 snapshot + leakage trap + structured missingness + subsampling
 - [x] `scripts/validate_v4_dataset.py` — full validation per `docs/v4/validation_spec.md`
 - [x] End-to-end: generate bundle → build CSV → validate → all 7 mandatory checks pass
-- [x] LR AUC 0.659 (without trap); 0.03+ boost with trap
+- [x] LR AUC 0.652 (without trap); 0.034 boost with trap
 
-### v4-M2: Documentation + release ⬜
+### v4-M2: Documentation + release ✓ (PR #22)
 
-- [ ] Generate `lead_scoring_intro_v4.csv` (in datasets-private repo)
-- [ ] Write `RELEASE_v4.md`
-- [ ] Update dataset repo README
-- [ ] Update `.agent-plan.md` to reflect completion
+- [x] Generate `lead_scoring_intro_v4.csv` (in datasets-private repo)
+- [x] Write `RELEASE_v4.md` with column dictionary, validation results, pedagogical notes
+- [x] Update dataset repo README (v4 now recommended, v3 superseded)
+- [x] Tune snapshot day (14 → 10) for leakage trap boost ≥ 0.03 after stacking fix
+- [x] Add StandardScaler to validation script LR for convergence
+- [x] Update `.agent-plan.md` to reflect completion
 
 ---
 
diff --git a/scripts/build_v4_snapshot.py b/scripts/build_v4_snapshot.py
index 7e228de..61ba1c8 100644
--- a/scripts/build_v4_snapshot.py
+++ b/scripts/build_v4_snapshot.py
@@ -5,7 +5,7 @@
     python scripts/build_v4_snapshot.py OUTPUT_CSV
 
 Produces a 1000-row × 18-column CSV at ~30% conversion rate with:
-- Day-14 windowed features
+- Day-10 windowed features
 - Structured missingness (MAR for web_sessions, seniority)
 - Leakage trap (total_touches_all using full 90-day data)
 - Stratified subsampling
@@ -27,7 +27,7 @@
 # ---------------------------------------------------------------------------
 SEED = 42
 N_LEADS = 5000
-SNAPSHOT_DAY = 14
+SNAPSHOT_DAY = 10
 SUBSAMPLE_N = 1000
 TARGET_RATE = 0.30
 
@@ -72,7 +72,7 @@
 
 
 def generate_bundle(seed: int = SEED, n_leads: int = N_LEADS) -> pd.DataFrame:
-    """Generate a full bundle and return the day-14 snapshot."""
+    """Generate a full bundle and return the day-10 snapshot."""
     gen = Generator.from_recipe(
         "b2b_saas_procurement_v1",
         seed=seed,
diff --git a/scripts/validate_v4_dataset.py b/scripts/validate_v4_dataset.py
index 09c60a4..c57cd95 100644
--- a/scripts/validate_v4_dataset.py
+++ b/scripts/validate_v4_dataset.py
@@ -16,7 +16,7 @@
 import pandas as pd
 from sklearn.linear_model import LogisticRegression
 from sklearn.metrics import roc_auc_score
-from sklearn.preprocessing import LabelEncoder
+from sklearn.preprocessing import LabelEncoder, StandardScaler
 
 # ---------------------------------------------------------------------------
 # Constants
@@ -120,9 +120,12 @@ def _fit_lr(df: pd.DataFrame, exclude_cols: list[str] | None = None) -> float:
     x_df = x_df.select_dtypes(include=[np.number])
     x_df = x_df.fillna(x_df.median())
 
+    scaler = StandardScaler()
+    x_scaled = scaler.fit_transform(x_df)
+
     lr = LogisticRegression(max_iter=2000, random_state=42)
-    lr.fit(x_df, y)
-    probs = lr.predict_proba(x_df)[:, 1]
+    lr.fit(x_scaled, y)
+    probs = lr.predict_proba(x_scaled)[:, 1]
     return float(roc_auc_score(y, probs))