Feature/llm cache redis semantic issue 362

api/app.py

@@ -55,11 +55,18 @@
     validation_router,
     ws_router,
     streaming_router,
+    llm_usage_router,
+    llm_cache_metrics_router,
     llm_router,
     reports_router,
     alerts_router,
 )
+
+
+
 from api.routers.monitoring import record_latency
+
+
 from api.routers.ws import poll_and_broadcast_transactions
 
 # Setup distributed tracing (issue #336)
@@ -167,12 +174,16 @@ async def _latency_middleware(request: Request, call_next):
 app.include_router(chat_router)
 app.include_router(ws_router)
 app.include_router(streaming_router)
+app.include_router(llm_usage_router)
+app.include_router(llm_cache_metrics_router)
+
 app.include_router(llm_router)
 app.include_router(reports_router)
 app.include_router(alerts_router)
 
 
 @app.get("/health", tags=["ops"])
+
 async def health():
     return {"status": "ok"}
 

api/routers/__init__.py

@@ -21,11 +21,15 @@
 from api.routers.validation import router as validation_router
 from api.routers.ws import router as ws_router
 from api.routers.streaming import router as streaming_router
+from api.routers.llm_usage import router as llm_usage_router
+from api.routers.llm_cache_metrics import router as llm_cache_metrics_router
+
 from api.routers.llm import router as llm_router
 from api.routers.reports import router as reports_router
 from api.routers.alerts import router as alerts_router
 
 __all__ = [
+
     "accounts_router",
     "audit_router",
     "backup_router",
@@ -48,7 +52,9 @@
     "validation_router",
     "ws_router",
     "streaming_router",
+    "llm_usage_router",
     "llm_router",
     "reports_router",
     "alerts_router",
 ]
+

api/routers/llm_cache_metrics.py

@@ -0,0 +1,19 @@
+"""LLM semantic cache metrics endpoints."""
+
+from __future__ import annotations
+
+from typing import Any, Dict
+
+from fastapi import APIRouter
+
+from astroml.cache.redis_cache import RedisCache
+from astroml.llm.llm_cached_client import get_semantic_cache_metrics
+
+router = APIRouter(prefix="/api/v1/llm", tags=["llm"])
+
+
+@router.get("/cache/semantic/metrics", response_model=Dict[str, Any])
+def semantic_cache_metrics():
+    """Return semantic cache hit/miss and avg lookup latency."""
+    return get_semantic_cache_metrics(redis_cache=RedisCache())
+

api/routers/llm_usage.py

@@ -0,0 +1,41 @@
+"""LLM usage and cost monitoring endpoints.
+
+These endpoints expose:
+- recent LLM call events (all calls logged)
+- rolling cost summaries
+
+Prometheus metrics are emitted by ``LLMUsageTracker``.
+"""
+
+from __future__ import annotations
+
+from typing import Any, Dict, List, Optional
+
+from fastapi import APIRouter, Query
+
+from astroml.tracking.llm_usage_tracker import default_llm_usage_tracker
+
+router = APIRouter(prefix="/api/v1/llm", tags=["llm"])
+
+
+@router.get("/usage/recent", response_model=List[Dict[str, Any]])
+def recent_llm_usage(limit: int = Query(100, ge=1, le=1000)):
+    """Return the most recent recorded LLM calls."""
+    return default_llm_usage_tracker.recent_calls(limit=limit)
+
+
+@router.get("/usage/summary", response_model=Dict[str, Any])
+def usage_summary():
+    """Return a lightweight summary based on recent in-memory buffer."""
+    events = default_llm_usage_tracker.recent_calls(limit=5000)
+    total_calls = len(events)
+    total_cost_usd = sum(float(e.get("cost_usd", 0.0) or 0.0) for e in events)
+    total_tokens = sum(int(e.get("total_tokens", 0) or 0) for e in events)
+
+    return {
+        "total_calls": total_calls,
+        "total_cost_usd": round(total_cost_usd, 6),
+        "total_tokens": total_tokens,
+        "window": "in-memory-recent (up to last 5000 events)",
+    }
+