feat(agents): enhance agent core with state management

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

backend/app/agents/state.py

@@ -138,6 +138,18 @@ class AgentState(TypedDict):
     memory_context: str | None
     current_datetime_context: str | None
     current_datetime_reference: dict[str, str] | None
+    runtime_request_context: dict[str, Any] | None
+    task_graph: dict[str, Any] | None
+    scheduled_subtasks: list[dict[str, Any]]
+    recalled_retrospectives: list[dict[str, Any]]
+    retrospective_shortlist: list[dict[str, Any]]
+    skill_shortlist: list[dict[str, Any]]
+    skill_activation_records: list[dict[str, Any]]
+    execution_decision: dict[str, Any] | None
+    merge_report: dict[str, Any] | None
+    verification_report: dict[str, Any] | None
+    feature_flags: dict[str, bool]
+    observability_report: dict[str, Any] | None
 
     turn_context: dict[str, Any] | None
     routing_decision: dict[str, Any] | None
@@ -254,6 +266,18 @@ def initial_state(user_id: str, conversation_id: str) -> AgentState:
         memory_context=None,
         current_datetime_context=None,
         current_datetime_reference=None,
+        runtime_request_context=None,
+        task_graph=None,
+        scheduled_subtasks=[],
+        recalled_retrospectives=[],
+        retrospective_shortlist=[],
+        skill_shortlist=[],
+        skill_activation_records=[],
+        execution_decision=None,
+        merge_report=None,
+        verification_report=None,
+        feature_flags={},
+        observability_report=None,
         turn_context=None,
         routing_decision=None,
         continuity_state=None,