X/misc fixes

agents-api/agents_api/app.py

@@ -22,15 +22,22 @@ class ObjectWithState(Protocol):
     state: State
 
 
+pool = None
+
+
 # TODO: This currently doesn't use env.py, we should move to using them
 @asynccontextmanager
 async def lifespan(*containers: FastAPI | ObjectWithState):
     # INIT POSTGRES #
     pg_dsn = os.environ.get("PG_DSN")
 
+    global pool
+    if not pool:
+        pool = await create_db_pool(pg_dsn)
+
     for container in containers:
         if hasattr(container, "state") and not getattr(container.state, "postgres_pool", None):
-            container.state.postgres_pool = await create_db_pool(pg_dsn)
+            container.state.postgres_pool = pool
 
     # INIT S3 #
     s3_access_key = os.environ.get("S3_ACCESS_KEY")
@@ -50,13 +57,13 @@ async def lifespan(*containers: FastAPI | ObjectWithState):
     try:
         yield
     finally:
-        # CLOSE POSTGRES #
-        for container in containers:
-            if hasattr(container, "state") and getattr(container.state, "postgres_pool", None):
-                pool = getattr(container.state, "postgres_pool", None)
-                if pool:
-                    await pool.close()
-                container.state.postgres_pool = None
+        # # CLOSE POSTGRES #
+        # for container in containers:
+        #     if hasattr(container, "state") and getattr(container.state, "postgres_pool", None):
+        #         pool = getattr(container.state, "postgres_pool", None)
+        #         if pool:
+        #             await pool.close()
+        #         container.state.postgres_pool = None
 
         # CLOSE S3 #
         for container in containers:

agents-api/agents_api/queries/docs/search_docs_by_embedding.py

@@ -13,7 +13,7 @@
 search_docs_by_embedding_query = """
 SELECT * FROM search_by_vector(
     $1, -- developer_id
-    $2::vector(1024), -- query_embedding
+    $2::vector(1024), -- embedding
     $3::text[], -- owner_types
     $4::uuid[], -- owner_ids
     $5, -- k
@@ -33,7 +33,7 @@
 async def search_docs_by_embedding(
     *,
     developer_id: UUID,
-    query_embedding: list[float],
+    embedding: list[float],
     k: int = 10,
     owners: list[tuple[Literal["user", "agent"], UUID]],
     confidence: float = 0.5,
@@ -44,7 +44,7 @@ async def search_docs_by_embedding(
 
     Parameters:
         developer_id (UUID): The ID of the developer.
-        query_embedding (List[float]): The vector to query.
+        embedding (List[float]): The vector to query.
         k (int): The number of results to return.
         owners (list[tuple[Literal["user", "agent"], UUID]]): List of (owner_type, owner_id) tuples.
         confidence (float): The confidence threshold for the search.
@@ -56,11 +56,11 @@ async def search_docs_by_embedding(
     if k < 1:
         raise HTTPException(status_code=400, detail="k must be >= 1")
 
-    if not query_embedding:
+    if not embedding:
         raise HTTPException(status_code=400, detail="Empty embedding provided")
 
-    # Convert query_embedding to a string
-    query_embedding_str = f"[{', '.join(map(str, query_embedding))}]"
+    # Convert embedding to a string
+    embedding_str = f"[{', '.join(map(str, embedding))}]"
 
     # Extract owner types and IDs
     owner_types: list[str] = [owner[0] for owner in owners]
@@ -70,7 +70,7 @@ async def search_docs_by_embedding(
         search_docs_by_embedding_query,
         [
             developer_id,
-            query_embedding_str,
+            embedding_str,
             owner_types,
             owner_ids,
             k,

agents-api/agents_api/queries/utils.py

@@ -116,7 +116,7 @@ async def wrapper(
             pool = (
                 connection_pool
                 if connection_pool is not None
-                else cast(asyncpg.Pool, app.state.postgres_pool)
+                else cast(asyncpg.Pool, getattr(app.state, "postgres_pool", None))
             )
 
             try:

agents-api/agents_api/routers/docs/search_docs.py

@@ -40,22 +40,22 @@ def get_search_fn_and_params(
             }
 
         case VectorDocSearchRequest(
-            vector=query_embedding,
+            vector=embedding,
             limit=k,
             confidence=confidence,
             metadata_filter=metadata_filter,
         ):
             search_fn = search_docs_by_embedding
             params = {
-                "query_embedding": query_embedding,
+                "embedding": embedding,
                 "k": k * 3 if search_params.mmr_strength > 0 else k,
                 "confidence": confidence,
                 "metadata_filter": metadata_filter,
             }
 
         case HybridDocSearchRequest(
             text=query,
-            vector=query_embedding,
+            vector=embedding,
             lang=lang,
             limit=k,
             confidence=confidence,
@@ -66,7 +66,7 @@ def get_search_fn_and_params(
             search_fn = search_docs_hybrid
             params = {
                 "text_query": query,
-                "embedding": query_embedding,
+                "embedding": embedding,
                 "k": k * 3 if search_params.mmr_strength > 0 else k,
                 "confidence": confidence,
                 "alpha": alpha,
@@ -111,7 +111,7 @@ async def search_user_docs(
         and len(docs) > search_params.limit
     ):
         indices = maximal_marginal_relevance(
-            np.asarray(params["query_embedding"]),
+            np.asarray(params["embedding"]),
             [doc.snippet.embedding for doc in docs],
             k=search_params.limit,
         )
@@ -160,7 +160,7 @@ async def search_agent_docs(
         and len(docs) > search_params.limit
     ):
         indices = maximal_marginal_relevance(
-            np.asarray(params["query_embedding"]),
+            np.asarray(params["embedding"]),
             [doc.snippet.embedding for doc in docs],
             k=search_params.limit,
         )

memory-store/migrations/000004_agents.up.sql

@@ -14,7 +14,7 @@ CREATE TABLE IF NOT EXISTS agents (
     ),
     about TEXT CONSTRAINT ct_agents_about_length CHECK (
         about IS NULL
-        OR length(about) <= 1000
+        OR length(about) <= 5000
     ),
     instructions TEXT[] DEFAULT ARRAY[]::TEXT[],
     model TEXT NOT NULL,

memory-store/migrations/000015_entries.up.sql

@@ -42,7 +42,7 @@ CREATE TABLE IF NOT EXISTS entries (
     token_count INTEGER DEFAULT NULL,
     tokenizer TEXT NOT NULL,
     created_at TIMESTAMPTZ NOT NULL DEFAULT CURRENT_TIMESTAMP,
-    timestamp DOUBLE PRECISION NOT NULL,
+    timestamp TIMESTAMPTZ NOT NULL,
     CONSTRAINT pk_entries PRIMARY KEY (session_id, entry_id, created_at),
     CONSTRAINT ct_content_is_array_of_objects CHECK (all_jsonb_elements_are_objects (content))
 );