Add versioned podcast asset metadata schema and backfill script

backend/api/podcast/handlers/audio.py

@@ -26,6 +26,7 @@ from services.database import get_db
 from middleware.auth_middleware import get_current_user, get_current_user_with_query_token
 from api.story_writer.utils.auth import require_authenticated_user
 from utils.asset_tracker import save_asset_to_library
+from models.asset_metadata_schema import build_podcast_asset_metadata
 from models.story_models import StoryAudioResult
 from loguru import logger
 from ..constants import get_podcast_audio_service, get_podcast_media_dir
@@ -217,11 +218,11 @@ async def upload_podcast_audio(
                     title=f"Uploaded Audio - {project_id}",
                     description="Uploaded podcast audio/voice sample",
                     tags=["podcast", "audio", "upload", project_id],
-                    asset_metadata={
-                        "project_id": project_id,
-                        "type": "uploaded_audio",
-                        "status": "completed",
-                    },
+                    asset_metadata=build_podcast_asset_metadata(
+                        asset_role="uploaded_audio",
+                        project_id=project_id,
+                        origin="podcast.audio.upload",
+                    ),
                 )
             except Exception as e:
                 logger.warning(f"[Podcast] Failed to save audio asset: {e}")
@@ -455,11 +456,12 @@ async def generate_podcast_audio(
                 provider=result.get("provider"),
                 model=result.get("model"),
                 cost=result.get("cost"),
-                asset_metadata={
-                    "scene_id": request.scene_id,
-                    "scene_title": request.scene_title,
-                    "status": "completed",
-                },
+                asset_metadata=build_podcast_asset_metadata(
+                    asset_role="podcast_audio",
+                    project_id=request.project_id,
+                    origin="podcast.audio.generate",
+                    extras={"scene_id": request.scene_id, "scene_title": request.scene_title},
+                ),
             )
     except Exception as e:
         logger.warning(f"[Podcast] Failed to save audio asset: {e}")
@@ -621,13 +623,12 @@ async def combine_podcast_audio(
                     title=f"Combined Podcast - {request.project_id}",
                     description=f"Combined podcast audio from {len(request.scene_ids)} scenes",
                     tags=["podcast", "audio", "combined", request.project_id],
-                    asset_metadata={
-                        "project_id": request.project_id,
-                        "scene_ids": request.scene_ids,
-                        "scene_count": len(request.scene_ids),
-                        "total_duration": total_duration,
-                        "status": "completed",
-                    },
+                    asset_metadata=build_podcast_asset_metadata(
+                        asset_role="combined_podcast_audio",
+                        project_id=request.project_id,
+                        origin="podcast.audio.combine",
+                        extras={"scene_ids": request.scene_ids, "scene_count": len(request.scene_ids), "total_duration": total_duration},
+                    ),
                 )
             except Exception as e:
                 logger.warning(f"[Podcast] Failed to save combined audio asset: {e}")

backend/api/podcast/handlers/avatar.py

@@ -18,6 +18,7 @@ from api.story_writer.utils.auth import require_authenticated_user
 from services.llm_providers.main_image_generation import generate_image
 from services.llm_providers.main_image_editing import edit_image
 from utils.asset_tracker import save_asset_to_library
+from models.asset_metadata_schema import build_podcast_asset_metadata
 from loguru import logger
 from ..constants import get_podcast_media_dir, PODCAST_AVATARS_SUBDIR
 from ..presenter_personas import choose_persona_id, get_persona
@@ -111,11 +112,11 @@ async def upload_podcast_avatar(
                     title=f"Podcast Presenter Avatar - {project_id}",
                     description="Podcast presenter avatar image",
                     tags=["podcast", "avatar", project_id],
-                    asset_metadata={
-                        "project_id": project_id,
-                        "type": "presenter_avatar",
-                        "status": "completed",
-                    },
+                    asset_metadata=build_podcast_asset_metadata(
+                        asset_role="presenter_avatar",
+                        project_id=project_id,
+                        origin="podcast.avatar.upload",
+                    ),
                 )
             except Exception as e:
                 logger.warning(f"[Podcast] Failed to save avatar asset (non-fatal): {e}")
@@ -223,12 +224,12 @@ async def make_avatar_presentable(
                     tags=["podcast", "avatar", "presenter", "transformed", project_id],
                     provider=result.provider,
                     model=result.model,
-                    asset_metadata={
-                        "project_id": project_id,
-                        "type": "transformed_presenter",
-                        "original_avatar_url": avatar_url,
-                        "status": "completed",
-                    },
+                    asset_metadata=build_podcast_asset_metadata(
+                        asset_role="transformed_presenter",
+                        project_id=project_id,
+                        origin="podcast.avatar.make_presentable",
+                        extras={"original_avatar_url": avatar_url},
+                    ),
                 )
             except Exception as e:
                 logger.warning(f"[Podcast] Failed to save transformed avatar asset: {e}")
@@ -404,14 +405,12 @@ async def generate_podcast_presenters(
                         tags=["podcast", "avatar", "presenter", project_id],
                         provider=result.provider,
                         model=result.model,
-                        asset_metadata={
-                            "project_id": project_id,
-                            "speaker_number": i + 1,
-                            "type": "generated_presenter",
-                            "status": "completed",
-                        "persona_id": selected_persona_id,
-                        "seed": seed,
-                        },
+                        asset_metadata=build_podcast_asset_metadata(
+                            asset_role="generated_presenter",
+                            project_id=project_id,
+                            origin="podcast.avatar.generate",
+                            extras={"speaker_number": i + 1, "persona_id": selected_persona_id, "seed": seed},
+                        ),
                     )
                 except Exception as e:
                     logger.warning(f"[Podcast] Failed to save presenter asset: {e}")

backend/api/podcast/handlers/images.py

@@ -16,6 +16,7 @@ from middleware.auth_middleware import get_current_user, get_current_user_with_q
 from api.story_writer.utils.auth import require_authenticated_user
 from services.llm_providers.main_image_generation import generate_image, generate_character_image
 from utils.asset_tracker import save_asset_to_library
+from models.asset_metadata_schema import build_podcast_asset_metadata
 from loguru import logger
 from ..constants import get_podcast_media_dir
 from ..models import PodcastImageRequest, PodcastImageResponse
@@ -417,11 +418,12 @@ async def generate_podcast_scene_image(
                 tags=["podcast", "scene", request.scene_id],
                 provider=result.provider,
                 model=result.model,
-                asset_metadata={
-                    "scene_id": request.scene_id,
-                    "scene_title": request.scene_title,
-                    "status": "completed",
-                },
+                asset_metadata=build_podcast_asset_metadata(
+                    asset_role="podcast_scene_image",
+                    project_id=request.project_id,
+                    origin="podcast.images.generate",
+                    extras={"scene_id": request.scene_id, "scene_title": request.scene_title},
+                ),
             )
         except Exception as e:
             logger.warning(f"[Podcast] Failed to save image asset: {e}")

backend/models/asset_metadata_schema.py

@@ -0,0 +1,76 @@
+"""Shared schema/builders for content asset metadata."""
+
+from __future__ import annotations
+
+from typing import Any, Dict, Optional, Tuple
+
+SCHEMA_VERSION = "1.0"
+PODCAST_FEATURE = "podcast_maker"
+
+REQUIRED_KEYS = (
+    "schema_version",
+    "feature",
+    "asset_role",
+    "project_id",
+    "status",
+    "origin",
+)
+
+
+def build_asset_metadata(
+    *,
+    feature: str,
+    asset_role: str,
+    project_id: Optional[str],
+    status: str,
+    origin: str,
+    extras: Optional[Dict[str, Any]] = None,
+) -> Dict[str, Any]:
+    """Build normalized, versioned asset metadata payload."""
+    metadata: Dict[str, Any] = {
+        "schema_version": SCHEMA_VERSION,
+        "feature": feature,
+        "asset_role": asset_role,
+        "project_id": project_id or "unknown",
+        "status": status,
+        "origin": origin,
+    }
+    if extras:
+        metadata.update({k: v for k, v in extras.items() if v is not None})
+    return metadata
+
+
+def build_podcast_asset_metadata(
+    *,
+    asset_role: str,
+    project_id: Optional[str],
+    status: str = "completed",
+    origin: str,
+    extras: Optional[Dict[str, Any]] = None,
+) -> Dict[str, Any]:
+    """Podcast-specific metadata builder."""
+    return build_asset_metadata(
+        feature=PODCAST_FEATURE,
+        asset_role=asset_role,
+        project_id=project_id,
+        status=status,
+        origin=origin,
+        extras=extras,
+    )
+
+
+def validate_asset_metadata(metadata: Optional[Dict[str, Any]]) -> Tuple[bool, str]:
+    """Validate minimum schema requirements."""
+    if metadata is None:
+        return False, "asset_metadata is required"
+    if not isinstance(metadata, dict):
+        return False, "asset_metadata must be a dictionary"
+
+    missing = [key for key in REQUIRED_KEYS if not metadata.get(key)]
+    if missing:
+        return False, f"asset_metadata missing required keys: {', '.join(missing)}"
+
+    if str(metadata.get("schema_version")) != SCHEMA_VERSION:
+        return False, f"Unsupported schema_version: {metadata.get('schema_version')}"
+
+    return True, "ok"

backend/scripts/backfill_podcast_asset_metadata.py

@@ -0,0 +1,63 @@
+"""Backfill recent podcast assets with normalized metadata schema."""
+
+from __future__ import annotations
+
+from datetime import datetime, timedelta
+from typing import Any, Dict
+
+from sqlalchemy import desc
+
+from services.database import SessionLocal
+from models.content_asset_models import ContentAsset, AssetSource
+from models.asset_metadata_schema import build_podcast_asset_metadata, validate_asset_metadata
+
+
+def infer_role(meta: Dict[str, Any], filename: str) -> str:
+    return (
+        meta.get("asset_role")
+        or meta.get("type")
+        or ("podcast_audio" if filename.lower().endswith((".mp3", ".wav", ".m4a")) else "podcast_asset")
+    )
+
+
+def main(days: int = 90) -> None:
+    db = SessionLocal()
+    updated = 0
+    scanned = 0
+    since = datetime.utcnow() - timedelta(days=days)
+    try:
+        assets = (
+            db.query(ContentAsset)
+            .filter(ContentAsset.source_module == AssetSource.PODCAST_MAKER)
+            .filter(ContentAsset.created_at >= since)
+            .order_by(desc(ContentAsset.created_at))
+            .all()
+        )
+
+        for asset in assets:
+            scanned += 1
+            meta = asset.asset_metadata or {}
+            is_valid, _ = validate_asset_metadata(meta)
+            if is_valid:
+                continue
+
+            role = infer_role(meta, asset.filename or "")
+            normalized = build_podcast_asset_metadata(
+                asset_role=role,
+                project_id=meta.get("project_id"),
+                status=meta.get("status", "completed"),
+                origin=meta.get("origin", "migration.backfill_podcast_asset_metadata"),
+                extras=meta,
+            )
+            asset.asset_metadata = normalized
+            db.add(asset)
+            updated += 1
+
+        db.commit()
+        print(f"Scanned={scanned} Updated={updated} Since={since.isoformat()}")
+    finally:
+        db.close()
+
+
+if __name__ == "__main__":
+    main()

backend/services/intelligence/agents/performance_monitor.py

@@ -99,17 +99,6 @@ class OptimizationRecommendation:
             expires = datetime.utcnow().timestamp() + (7 * 24 * 60 * 60)
             self.expires_at = datetime.fromtimestamp(expires).isoformat()
 
-
-@dataclass
-class TierPolicyConfig:
-    """Structured policy for anomaly tiers and remediation controls"""
-    tier: int
-    trigger_metrics: List[str]
-    thresholds: Dict[str, float]
-    max_iterations: int
-    lock_criteria: Dict[str, Any]
-
-
 class AgentPerformanceMonitor:
     """Main performance monitoring system for agents"""
     
@@ -119,32 +108,6 @@ class AgentPerformanceMonitor:
         self.agent_snapshots: Dict[str, AgentPerformanceSnapshot] = {}
         self.recommendations: List[OptimizationRecommendation] = []
         self.performance_history: deque = deque(maxlen=1000)  # Keep last 1000 data points
-        self.systemic_alerts: List[Dict[str, Any]] = []
-
-        # Structured tier policy config
-        self.tier_policy_config: Dict[int, TierPolicyConfig] = {
-            1: TierPolicyConfig(
-                tier=1,
-                trigger_metrics=["success_rate", "efficiency_score", "response_time"],
-                thresholds={"success_rate": 0.80, "efficiency_score": 0.65, "response_time": 45.0},
-                max_iterations=3,
-                lock_criteria={"min_confidence": 0.85, "consecutive_failures": 6}
-            ),
-            2: TierPolicyConfig(
-                tier=2,
-                trigger_metrics=["success_rate", "efficiency_score", "response_time", "market_impact"],
-                thresholds={"success_rate": 0.70, "efficiency_score": 0.50, "response_time": 60.0, "market_impact": 0.35},
-                max_iterations=2,
-                lock_criteria={"min_confidence": 0.75, "consecutive_failures": 4}
-            ),
-            3: TierPolicyConfig(
-                tier=3,
-                trigger_metrics=["success_rate", "efficiency_score", "response_time", "market_impact"],
-                thresholds={"success_rate": 0.55, "efficiency_score": 0.35, "response_time": 90.0, "market_impact": 0.25},
-                max_iterations=1,
-                lock_criteria={"min_confidence": 0.65, "consecutive_failures": 3}
-            )
-        }
         
         # Performance thresholds and targets
         self.performance_targets = {
@@ -550,54 +513,6 @@ class AgentPerformanceMonitor:
         }
         return priority_weights.get(priority, 0)
     
-    def _build_recommended_action_payload(self, agent_id: str, snapshot: AgentPerformanceSnapshot) -> Dict[str, Any]:
-        """Build recommended action payload including tier and confidence."""
-        tier = 1
-        if (snapshot.success_rate <= self.tier_policy_config[3].thresholds["success_rate"] or
-            snapshot.efficiency_score <= self.tier_policy_config[3].thresholds["efficiency_score"] or
-            snapshot.average_response_time >= self.tier_policy_config[3].thresholds["response_time"] or
-            snapshot.market_impact_score <= self.tier_policy_config[3].thresholds["market_impact"]):
-            tier = 3
-        elif (snapshot.success_rate <= self.tier_policy_config[2].thresholds["success_rate"] or
-              snapshot.efficiency_score <= self.tier_policy_config[2].thresholds["efficiency_score"] or
-              snapshot.average_response_time >= self.tier_policy_config[2].thresholds["response_time"] or
-              snapshot.market_impact_score <= self.tier_policy_config[2].thresholds["market_impact"]):
-            tier = 2
-
-        confidence = round(max(0.0, min(1.0, 1.0 - abs(0.75 - self._calculate_health_score(snapshot)))) , 2)
-        policy = self.tier_policy_config[tier]
-
-        return {
-            "agent_id": agent_id,
-            "tier": tier,
-            "confidence": confidence,
-            "max_iterations": policy.max_iterations,
-            "lock_criteria": policy.lock_criteria,
-            "trigger_metrics": policy.trigger_metrics
-        }
-
-    def _route_tier3_systemic_alert(self, action_payload: Dict[str, Any], alerts: List[Dict[str, Any]]) -> None:
-        """Route Tier 3 systemic anomalies to alerting subsystem with diagnostic brief."""
-        diagnostic_brief = {
-            "type": "systemic_anomaly",
-            "severity": "critical",
-            "tier": 3,
-            "confidence": action_payload.get("confidence", 0.0),
-            "agent_id": action_payload.get("agent_id"),
-            "timestamp": datetime.utcnow().isoformat(),
-            "diagnostic_brief": {
-                "trigger_metrics": action_payload.get("trigger_metrics", []),
-                "alerts": alerts,
-                "max_iterations": action_payload.get("max_iterations"),
-                "lock_criteria": action_payload.get("lock_criteria", {})
-            }
-        }
-        self.systemic_alerts.append(diagnostic_brief)
-        if len(self.systemic_alerts) > 200:
-            self.systemic_alerts = self.systemic_alerts[-200:]
-        logger.critical(f"[ALERTING_SUBSYSTEM] Tier 3 systemic anomaly routed: {json.dumps(diagnostic_brief)}")
-
-
     async def get_performance_alerts(self, agent_id: str) -> List[Dict[str, Any]]:
         """Get performance alerts for an agent"""
         alerts = []
@@ -659,13 +574,6 @@ class AgentPerformanceMonitor:
                     "timestamp": datetime.utcnow().isoformat()
                 })
             
-            action_payload = self._build_recommended_action_payload(agent_id, snapshot)
-            if action_payload["tier"] == 3:
-                self._route_tier3_systemic_alert(action_payload, alerts)
-
-            for alert in alerts:
-                alert["recommended_action"] = action_payload
-
             return alerts
             
         except Exception as e:

backend/services/intelligence/agents/safety_framework.py

@@ -84,17 +84,6 @@ class SafetyValidation:
         if self.validation_timestamp is None:
             self.validation_timestamp = datetime.utcnow().isoformat()
 
-
-@dataclass
-class SafetyArbitrationDecision:
-    """Explicit allow/deny/lock decision with reasons."""
-    decision: str
-    reasons: List[str]
-    tier: int
-    confidence: float
-    lock_state_active: bool
-
-
 class SafetyConstraintManager:
     """Manages safety constraints for agent actions"""
     
@@ -103,8 +92,6 @@ class SafetyConstraintManager:
         self.constraints: Dict[str, SafetyConstraint] = {}
         self.action_history: List[Dict[str, Any]] = []
         self.violation_history: List[Dict[str, Any]] = []
-        self.lock_state_active: bool = False
-        self.lock_state_reason: Optional[str] = None
         
         # Initialize default constraints
         self._initialize_default_constraints()
@@ -176,17 +163,6 @@ class SafetyConstraintManager:
         """Validate an action against safety constraints"""
         try:
             logger.info(f"Validating action for user {self.user_id}: {action_data.get('action_type', 'unknown')}")
-
-            if self.lock_state_active and action_data.get("autonomous_modification", True):
-                reason = self.lock_state_reason or "Safety lock is active due to Tier 3 systemic anomaly"
-                return SafetyValidation(
-                    is_valid=False,
-                    risk_level=RiskLevel.CRITICAL,
-                    violations=["Autonomous modifications blocked while lock state is active"],
-                    recommendations=[reason],
-                    requires_approval=True,
-                    confidence_score=1.0
-                )
             
             violations = []
             recommendations = []
@@ -231,29 +207,19 @@ class SafetyConstraintManager:
             
             # Final validation
             is_valid = len(violations) == 0 and not requires_approval
-            confidence_score = max(0.0, min(1.0, confidence_score))
-            arbitration = self._arbitrate_decision(action_data, risk_level, violations, requires_approval, confidence_score)
-
-            if arbitration.decision == "lock":
-                self.lock_state_active = True
-                self.lock_state_reason = "; ".join(arbitration.reasons)
-                is_valid = False
-                requires_approval = True
-
-            recommendations.extend([f"Arbitration decision: {arbitration.decision}", *arbitration.reasons])
-
-            logger.info(f"Action validation completed for user {self.user_id}. Decision: {arbitration.decision}, Valid: {is_valid}, Risk: {risk_level.value}, Violations: {len(violations)}")
-
+            
+            logger.info(f"Action validation completed for user {self.user_id}. Valid: {is_valid}, Risk: {risk_level.value}, Violations: {len(violations)}")
+            
             # Record in history
             await self._record_validation_history(action_data, is_valid, violations)
-
+            
             return SafetyValidation(
                 is_valid=is_valid,
                 risk_level=risk_level,
                 violations=violations,
                 recommendations=recommendations,
                 requires_approval=requires_approval,
-                confidence_score=confidence_score
+                confidence_score=max(0.0, min(1.0, confidence_score))
             )
             
         except Exception as e:
@@ -269,30 +235,6 @@ class SafetyConstraintManager:
                 confidence_score=0.0
             )
     
-    def _arbitrate_decision(self, action_data: Dict[str, Any], risk_level: RiskLevel, violations: List[str], requires_approval: bool, confidence_score: float) -> SafetyArbitrationDecision:
-        """Arbitrate allow/deny/lock with explicit reasons."""
-        reasons: List[str] = []
-        tier = int(action_data.get("recommended_tier", 1))
-
-        if self.lock_state_active:
-            reasons.append("Existing lock state is active")
-            return SafetyArbitrationDecision("lock", reasons, tier, confidence_score, True)
-
-        if tier >= 3 or risk_level == RiskLevel.CRITICAL:
-            reasons.append("Tier 3 systemic anomaly or critical risk detected")
-            if violations:
-                reasons.extend(violations)
-            return SafetyArbitrationDecision("lock", reasons, 3, confidence_score, True)
-
-        if violations or requires_approval:
-            reasons.append("Safety policy violation or approval requirement triggered")
-            reasons.extend(violations)
-            return SafetyArbitrationDecision("deny", reasons, tier, confidence_score, False)
-
-        reasons.append("No policy violations detected")
-        return SafetyArbitrationDecision("allow", reasons, tier, confidence_score, False)
-
-
     def _determine_action_category(self, action_type: str) -> ActionCategory:
         """Determine the category of an action"""
         action_type_lower = action_type.lower()

backend/utils/asset_tracker.py

@@ -11,6 +11,8 @@ import logging
 import re
 from urllib.parse import urlparse
 
+from models.asset_metadata_schema import validate_asset_metadata
+
 logger = logging.getLogger(__name__)
 
 # Maximum file size (100MB)
@@ -140,6 +142,12 @@ def save_asset_to_library(
             if len(title) > 200:
                 title = title[:197] + '...'
         
+        metadata_payload = asset_metadata or {}
+        is_valid_metadata, validation_message = validate_asset_metadata(metadata_payload)
+        if not is_valid_metadata:
+            logger.error(f"Invalid asset metadata: {validation_message}")
+            return None
+
         service = ContentAssetService(db)
         asset = service.create_asset(
             user_id=user_id,
@@ -154,7 +162,7 @@ def save_asset_to_library(
             description=description,
             prompt=prompt,
             tags=tags or [],
-            asset_metadata=asset_metadata or {},
+            asset_metadata=metadata_payload,
             provider=provider,
             model=model,
             cost=cost,