Implement CYA-WP-0008 llm-connect adapter integration.

Wire LLMConnectAdapter behind the existing LLMAdapter seam with config-driven
selection, graceful degradation, --offline mode, and bounded session context.
Add unit tests, integration docs, and update README/SCOPE/AGENTS.

src/cya/orchestrator.py

@@ -48,7 +48,9 @@ from cya.memory.reflections import (
     session_provenance,
 )
 from cya.safety.risk import classify, get_user_confirmation
-from cya.llm.adapter import AssistanceRequest, FakeLLMAdapter
+from cya.config import bound_session_turns
+from cya.llm.adapter import AssistanceRequest
+from cya.llm.factory import get_adapter
 
 
 console = Console()
@@ -59,6 +61,8 @@ def handle_request(
     *,
     explain_context: bool = False,
     dry_run: bool = False,
+    offline: bool = False,
+    session_turns: list[dict[str, str]] | None = None,
 ) -> None:
     """Primary orchestrator entry point.
 
@@ -158,10 +162,12 @@ def handle_request(
         console.print("[green]--dry-run acknowledged.[/green] No side-effects.")
         return
 
-    # 3. Call through the single LLMAdapter boundary (T04)
-    adapter = FakeLLMAdapter()
+    # 3. Call through the single LLMAdapter boundary (T04 / CYA-WP-0008)
+    adapter = get_adapter(offline=offline)
     ctx = (envelope.to_dict() if envelope else {}) or {}
     ctx["memory"] = memory  # T03: memory now in context passed to LLM (for personalization + explain)
+    if session_turns:
+        ctx["session_turns"] = bound_session_turns(session_turns)
     llm_request = AssistanceRequest(
         user_request=user_request,
         context=ctx,