feat(learning): raise writer lessons cap 15 → 60 (config.WRITER_LESSONS_LIMIT) — closes #157

עכשיו שהלקחים מנוכי-כפילות (סינתזה #158) ומאושרי-יו"ר, התקרה 15 חתכה בשקט הנחיות-סגנון
מובחנות ו-vetted (רישוי/סגנון 29>15). מעלים ל-60 (env-tunable WRITER_LESSONS_LIMIT) —
מכסה כל shard נוכחי עם מרווח; הסינתזה שומרת על אי-רדודנטיות בצמיחה. ה-WARN על חיתוך נשאר
רשת-הביטחון (חוקה §6) ומפנה לסינתזה/העלאת-התקרה. block_writer צורך דרך config.

Invariants: INV-LRN1/G10 (עדיין רק approved זורם; superseded מוחרג), G2. closes #157 בפועל.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

mcp-server/src/legal_mcp/config.py

@@ -293,6 +293,13 @@ LESSON_SYNTH_DRIFT_FLOOR = float(os.environ.get("LESSON_SYNTH_DRIFT_FLOOR", "0.8
 # quality backstop below this.
 LESSON_SYNTH_CLUSTER_THRESHOLD = float(os.environ.get("LESSON_SYNTH_CLUSTER_THRESHOLD", "0.68"))
 
+# How many CHAIR-APPROVED decision_lessons flow to the writer per practice_area
+# (get_recent_decision_lessons). Raised 15 → 60 (#158/#157): the lessons are now both
+# deduplicated (synthesis merges true near-duplicates) AND chair-approved, so the old
+# cap silently truncated distinct, vetted style guidance. 60 covers every current shard
+# with headroom; synthesis keeps the set non-redundant as it grows. env-tunable.
+WRITER_LESSONS_LIMIT = int(os.environ.get("WRITER_LESSONS_LIMIT", "60"))
+
 # Mistral OCR (fallback for scanned PDFs — replaces Google Cloud Vision)
 MISTRAL_API_KEY = os.environ.get("MISTRAL_API_KEY", "")
 

mcp-server/src/legal_mcp/services/block_writer.py

@@ -1028,7 +1028,8 @@ async def _build_style_context(practice_area: str = "") -> str:
     except Exception as e:
         logger.warning("methodology overrides not loaded: %s", e)
     try:
-        lessons = await db.get_recent_decision_lessons(limit=15, practice_area=practice_area)
+        lessons = await db.get_recent_decision_lessons(
+            limit=config.WRITER_LESSONS_LIMIT, practice_area=practice_area)
         if lessons:
             learned.append("\n**לקחים מהחלטות קודמות (decision_lessons):**")
             for ls in lessons:

mcp-server/src/legal_mcp/services/db.py

@@ -3147,19 +3147,18 @@ async def get_style_distance_history() -> list[dict]:
     ]
 
 
-async def get_recent_decision_lessons(limit: int = 15, practice_area: str = "") -> list[dict]:
+async def get_recent_decision_lessons(limit: int = 60, practice_area: str = "") -> list[dict]:
     """Per-decision learnings the chair/curator attached in /training (decision_lessons),
     so the writer consumes them too (T15). Prefers style/structure/lexicon, recent first.
 
-    Gate (INV-LRN1/G10): returns only CHAIR-APPROVED lessons (review_status='approved').
+    Gate (INV-LRN1/G10): returns only CHAIR-APPROVED lessons (review_status='approved');
-    Panel-written proposals stay out of the writer's context until the chair approves
+    superseded sources (merged by synthesis, #158) are excluded automatically.
-    them in /training — the chair's review is the gate, not the panel's 2/2 vote.
 
-    No silent cap (חוקה §6): when more approved lessons match than `limit`, the
+    Cap (#157/#158): callers pass config.WRITER_LESSONS_LIMIT (60). Raised from 15 once
-    overflow is dropped by ORDER BY created_at DESC — log it so coverage loss is
+    lessons became deduplicated (synthesis) AND chair-approved — the old cap silently
-    observable, not invisible. The real fix is lesson synthesis (TaskMaster #158):
+    truncated distinct, vetted style guidance. If even 60 is exceeded the overflow is
-    consolidate overlapping lessons into fewer, richer ones so the set fits without
+    dropped (ORDER BY created_at DESC) and logged (no silent cap, חוקה §6); that WARN is
-    truncation. Until then this WARN is the audit trail.
+    the signal to run synthesis (lesson_synthesize_pending) or raise WRITER_LESSONS_LIMIT.
     """
     pool = await get_pool()
     async with pool.acquire() as conn:
@@ -3185,7 +3184,8 @@ async def get_recent_decision_lessons(limit: int = 15, practice_area: str = "")
             if total > limit:
                 logger.warning(
                     "get_recent_decision_lessons: capped %d→%d approved lessons "
-                    "(practice_area=%r) — %d not reaching the writer. See TaskMaster #158 (synthesis).",
+                    "(practice_area=%r) — %d not reaching the writer. Run synthesis "
+                    "(lesson_synthesize_pending) or raise WRITER_LESSONS_LIMIT.",
                     total, limit, practice_area or "*", total - limit,
                 )
     return [dict(r) for r in rows]