@hazeljs/cli 0.5.2 → 0.5.4

package/@template-ai-native/package.json

@@ -1,6 +1,6 @@
 {
   "name": "hazeljs-ai-native-app",
-  "version": "0.5.2",
+  "version": "0.5.4",
   "description": "A HazelJS AI-native application with agents, RAG, and chat",
   "main": "dist/index.js",
   "scripts": {

package/@template-ai-native/prisma/schema.prisma

@@ -11,12 +11,12 @@ datasource db {
 }
 
 model Document {
-  id        String   @id @default(cuid())
+  id        String                     @id @default(cuid())
   content   String
   metadata  Json?
-  embedding Float[]  // Vector embedding for RAG
-  createdAt DateTime @default(now())
-  updatedAt DateTime @updatedAt
+  embedding Unsupported("vector(1536)")  // pgvector type for embeddings
+  createdAt DateTime                   @default(now())
+  updatedAt DateTime                   @updatedAt
 
   @@map("documents")
   // Vector indexes will be created by Prisma when pgvector extension is available

package/@template-ai-native/prisma/seed.ts

@@ -67,14 +67,18 @@ async function main() {
       // Generate embedding for the document
       const embedding = await embeddings.embed(doc.content);
       
-      // Create document with embedding
-      await prisma.document.create({
-        data: {
-          content: doc.content,
-          metadata: doc.metadata,
-          embedding: embedding,
-        },
-      });
+      // Use raw SQL to insert document with vector embedding
+      await prisma.$executeRaw`
+        INSERT INTO documents (id, content, metadata, embedding, "createdAt", "updatedAt")
+        VALUES (
+          gen_random_uuid()::text,
+          ${doc.content},
+          ${JSON.stringify(doc.metadata)}::jsonb,
+          ${embedding}::vector,
+          NOW(),
+          NOW()
+        )
+      `;
       
       console.log(`✅ Created document: ${doc.metadata.source}`);
     } catch (error) {

package/@template-ai-native/src/rag/rag.controller.ts

@@ -65,18 +65,17 @@ export class RAGService {
     const queryVector = await this.embeddings.embed(query);
     
     // Perform vector similarity search using raw SQL for cosine similarity
+    // Note: We don't select the embedding column to avoid deserialization issues
     const results = await this.prisma.$queryRaw<Array<{
       id: string;
       content: string;
       metadata: any;
-      embedding: number[];
       similarity: number;
     }>>`
       SELECT 
         id,
         content,
         metadata,
-        embedding,
         (1 - (embedding <=> ${queryVector}::vector)) as similarity
       FROM documents
       ORDER BY embedding <=> ${queryVector}::vector

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hazeljs/cli",
-  "version": "0.5.2",
+  "version": "0.5.4",
   "description": "Command-line interface for scaffolding and generating HazelJS applications and components",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",
@@ -70,5 +70,5 @@
     "type": "opencollective",
     "url": "https://opencollective.com/hazeljs"
   },
-  "gitHead": "5904fb3c12954d1e796529de0ec735ec8e410787"
+  "gitHead": "92397b089e9afd5b985f10c365861891dcd700b8"
 }