@retrivora-ai/rag-engine 0.3.2 → 0.3.4

package/dist/{RagConfig-Crz-l02S.d.mts → RagConfig-Bpp39-um.d.mts}

@@ -20,12 +20,23 @@ interface ChatResponse {
     sources: VectorMatch[];
 }
 
+/**
+ * constants.ts — Centralized definitions of acceptable configuration values.
+ *
+ * This file serves as the single source of truth for all statically supported
+ * providers and stylistic options in the Retrivora AI RAG Engine.
+ */
+declare const VECTOR_DB_PROVIDERS: readonly ["pinecone", "pgvector", "postgresql", "mongodb", "milvus", "qdrant", "chromadb", "redis", "weaviate", "rest", "universal_rest", "custom"];
+declare const LLM_PROVIDERS: readonly ["openai", "anthropic", "ollama", "gemini", "rest", "universal_rest", "custom"];
+declare const EMBEDDING_PROVIDERS: readonly ["openai", "ollama", "gemini", "rest", "universal_rest", "custom"];
+
 /**
  * Master configuration interface for Retrivora AI.
  * Each consuming project provides one RagConfig object to drive
  * vector DB selection, LLM selection, embedding, and UI branding.
  */
-type VectorDBProvider = 'pinecone' | 'pgvector' | 'postgresql' | 'mongodb' | 'milvus' | 'qdrant' | 'chromadb' | 'redis' | 'weaviate' | 'rest' | 'universal_rest' | 'custom';
+
+type VectorDBProvider = typeof VECTOR_DB_PROVIDERS[number];
 interface VectorDBConfig {
     /** Which vector database to use */
     provider: VectorDBProvider;
@@ -51,7 +62,7 @@ interface VectorDBConfig {
      */
     options: Record<string, unknown>;
 }
-type LLMProvider = 'openai' | 'anthropic' | 'ollama' | 'gemini' | 'rest' | 'universal_rest' | 'custom';
+type LLMProvider = typeof LLM_PROVIDERS[number];
 interface LLMConfig {
     /** Which LLM provider to use */
     provider: LLMProvider;
@@ -77,7 +88,7 @@ interface LLMConfig {
      */
     options?: Record<string, unknown>;
 }
-type EmbeddingProvider = 'openai' | 'ollama' | 'gemini' | 'rest' | 'universal_rest' | 'custom';
+type EmbeddingProvider = typeof EMBEDDING_PROVIDERS[number];
 interface EmbeddingConfig {
     /** Which embedding provider to use */
     provider: EmbeddingProvider;

package/dist/{RagConfig-Crz-l02S.d.ts → RagConfig-Bpp39-um.d.ts}

@@ -20,12 +20,23 @@ interface ChatResponse {
     sources: VectorMatch[];
 }
 
+/**
+ * constants.ts — Centralized definitions of acceptable configuration values.
+ *
+ * This file serves as the single source of truth for all statically supported
+ * providers and stylistic options in the Retrivora AI RAG Engine.
+ */
+declare const VECTOR_DB_PROVIDERS: readonly ["pinecone", "pgvector", "postgresql", "mongodb", "milvus", "qdrant", "chromadb", "redis", "weaviate", "rest", "universal_rest", "custom"];
+declare const LLM_PROVIDERS: readonly ["openai", "anthropic", "ollama", "gemini", "rest", "universal_rest", "custom"];
+declare const EMBEDDING_PROVIDERS: readonly ["openai", "ollama", "gemini", "rest", "universal_rest", "custom"];
+
 /**
  * Master configuration interface for Retrivora AI.
  * Each consuming project provides one RagConfig object to drive
  * vector DB selection, LLM selection, embedding, and UI branding.
  */
-type VectorDBProvider = 'pinecone' | 'pgvector' | 'postgresql' | 'mongodb' | 'milvus' | 'qdrant' | 'chromadb' | 'redis' | 'weaviate' | 'rest' | 'universal_rest' | 'custom';
+
+type VectorDBProvider = typeof VECTOR_DB_PROVIDERS[number];
 interface VectorDBConfig {
     /** Which vector database to use */
     provider: VectorDBProvider;
@@ -51,7 +62,7 @@ interface VectorDBConfig {
      */
     options: Record<string, unknown>;
 }
-type LLMProvider = 'openai' | 'anthropic' | 'ollama' | 'gemini' | 'rest' | 'universal_rest' | 'custom';
+type LLMProvider = typeof LLM_PROVIDERS[number];
 interface LLMConfig {
     /** Which LLM provider to use */
     provider: LLMProvider;
@@ -77,7 +88,7 @@ interface LLMConfig {
      */
     options?: Record<string, unknown>;
 }
-type EmbeddingProvider = 'openai' | 'ollama' | 'gemini' | 'rest' | 'universal_rest' | 'custom';
+type EmbeddingProvider = typeof EMBEDDING_PROVIDERS[number];
 interface EmbeddingConfig {
     /** Which embedding provider to use */
     provider: EmbeddingProvider;

package/dist/{chunk-7BPQDPOC.mjs → chunk-GE4AVGA5.mjs}

@@ -11,7 +11,7 @@ import {
 // src/handlers/index.ts
 import { NextResponse } from "next/server";
 
-// src/config/serverConfig.ts
+// src/config/constants.ts
 var VECTOR_DB_PROVIDERS = [
   "pinecone",
   "pgvector",
@@ -26,8 +26,34 @@ var VECTOR_DB_PROVIDERS = [
   "universal_rest",
   "custom"
 ];
-var LLM_PROVIDERS = ["openai", "anthropic", "ollama", "rest", "universal_rest", "custom"];
-var EMBEDDING_PROVIDERS = ["openai", "ollama", "rest", "universal_rest", "custom"];
+var LLM_PROVIDERS = [
+  "openai",
+  "anthropic",
+  "ollama",
+  "gemini",
+  "rest",
+  "universal_rest",
+  "custom"
+];
+var EMBEDDING_PROVIDERS = [
+  "openai",
+  "ollama",
+  "gemini",
+  "rest",
+  "universal_rest",
+  "custom"
+];
+var PROVIDERS_WITH_EMBEDDINGS = [
+  "openai",
+  "ollama",
+  "gemini",
+  "rest",
+  "universal_rest"
+];
+var UI_VISUAL_STYLES = ["glass", "solid"];
+var UI_BORDER_RADIUS_OPTIONS = ["none", "sm", "md", "lg", "xl", "full"];
+
+// src/config/serverConfig.ts
 function readString(env, name) {
   var _a;
   const value = (_a = env[name]) == null ? void 0 : _a.trim();
@@ -94,6 +120,7 @@ function getRagConfig(env = process.env) {
   const llmApiKeyByProvider = {
     openai: readString(env, "OPENAI_API_KEY"),
     anthropic: readString(env, "ANTHROPIC_API_KEY"),
+    gemini: readString(env, "GEMINI_API_KEY"),
     ollama: void 0,
     universal_rest: readString(env, "LLM_API_KEY"),
     rest: readString(env, "LLM_API_KEY"),
@@ -101,6 +128,7 @@ function getRagConfig(env = process.env) {
   };
   const embeddingApiKeyByProvider = {
     openai: readString(env, "OPENAI_API_KEY"),
+    gemini: readString(env, "GEMINI_API_KEY"),
     ollama: void 0,
     universal_rest: (_q = readString(env, "EMBEDDING_API_KEY")) != null ? _q : readString(env, "OPENAI_API_KEY"),
     custom: (_r = readString(env, "EMBEDDING_API_KEY")) != null ? _r : readString(env, "OPENAI_API_KEY")
@@ -629,21 +657,19 @@ var ConfigValidator = class {
       }
     }
     if (config.borderRadius) {
-      const validValues = ["none", "sm", "md", "lg", "xl", "full"];
-      if (!validValues.includes(config.borderRadius)) {
+      if (!UI_BORDER_RADIUS_OPTIONS.includes(config.borderRadius)) {
         errors.push({
           field: "ui.borderRadius",
-          message: `borderRadius must be one of: ${validValues.join(", ")}`,
+          message: `borderRadius must be one of: ${UI_BORDER_RADIUS_OPTIONS.join(", ")}`,
           severity: "warning"
         });
       }
     }
     if (config.visualStyle) {
-      const validValues = ["glass", "solid"];
-      if (!validValues.includes(config.visualStyle)) {
+      if (!UI_VISUAL_STYLES.includes(config.visualStyle)) {
         errors.push({
           field: "ui.visualStyle",
-          message: `visualStyle must be one of: ${validValues.join(", ")}`,
+          message: `visualStyle must be one of: ${UI_VISUAL_STYLES.join(", ")}`,
           severity: "warning"
         });
       }
@@ -983,8 +1009,18 @@ var GeminiProvider = class {
   constructor(llmConfig, embeddingConfig) {
     if (!llmConfig.apiKey) throw new Error("[GeminiProvider] llmConfig.apiKey is required");
     this.client = new GoogleGenAI({ apiKey: llmConfig.apiKey });
-    this.llmConfig = llmConfig;
-    this.embeddingConfig = embeddingConfig;
+    this.llmConfig = __spreadProps(__spreadValues({}, llmConfig), {
+      model: this.sanitizeModel(llmConfig.model)
+    });
+    if (embeddingConfig) {
+      this.embeddingConfig = __spreadProps(__spreadValues({}, embeddingConfig), {
+        model: this.sanitizeModel(embeddingConfig.model)
+      });
+    }
+  }
+  sanitizeModel(model) {
+    if (!model) return model;
+    return model.split(":")[0];
   }
   async chat(messages, context, options) {
     var _a, _b, _c, _d, _e, _f;
@@ -1013,8 +1049,10 @@ ${context}`;
     return (_f = response.text) != null ? _f : "";
   }
   async embed(text, options) {
-    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j;
-    const model = (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004";
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k;
+    const model = this.sanitizeModel(
+      (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004"
+    );
     const apiKey = (_e = (_d = this.embeddingConfig) == null ? void 0 : _d.apiKey) != null ? _e : this.llmConfig.apiKey;
     const client = apiKey !== this.llmConfig.apiKey ? new GoogleGenAI({ apiKey }) : this.client;
     let content = text;
@@ -1031,14 +1069,20 @@ ${context}`;
     }
     const response = await client.models.embedContent({
       model,
-      contents: content
+      contents: content,
+      config: ((_h = this.embeddingConfig) == null ? void 0 : _h.dimensions) ? { outputDimensionality: this.embeddingConfig.dimensions } : void 0
+    }).catch((err) => {
+      console.error(`[GeminiProvider] Embedding failed for model "${model}":`, err.message);
+      throw err;
     });
-    return (_j = (_i = (_h = response.embeddings) == null ? void 0 : _h[0]) == null ? void 0 : _i.values) != null ? _j : [];
+    return (_k = (_j = (_i = response.embeddings) == null ? void 0 : _i[0]) == null ? void 0 : _j.values) != null ? _k : [];
   }
   async batchEmbed(texts, options) {
-    var _a, _b, _c, _d, _e, _f, _g, _h, _i;
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j;
     const vectors = [];
-    const model = (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004";
+    const model = this.sanitizeModel(
+      (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004"
+    );
     const apiKey = (_e = (_d = this.embeddingConfig) == null ? void 0 : _d.apiKey) != null ? _e : this.llmConfig.apiKey;
     const client = apiKey !== this.llmConfig.apiKey ? new GoogleGenAI({ apiKey }) : this.client;
     let contents = texts;
@@ -1051,12 +1095,16 @@ ${context}`;
     }
     const response = await client.models.embedContent({
       model,
-      contents
+      contents,
+      config: ((_h = this.embeddingConfig) == null ? void 0 : _h.dimensions) ? { outputDimensionality: this.embeddingConfig.dimensions } : void 0
+    }).catch((err) => {
+      console.error(`[GeminiProvider] Batch embedding failed for model "${model}":`, err.message);
+      throw err;
     });
-    return (_i = (_h = response.embeddings) == null ? void 0 : _h.map((e) => {
+    return (_j = (_i = response.embeddings) == null ? void 0 : _i.map((e) => {
       var _a2;
       return (_a2 = e.values) != null ? _a2 : [];
-    })) != null ? _i : [];
+    })) != null ? _j : [];
   }
   async ping() {
     try {
@@ -1661,8 +1709,7 @@ var EmbeddingStrategyResolver = class {
    * Check if an LLM provider natively supports embeddings
    */
   static supportsEmbedding(provider) {
-    const providersWithEmbedding = ["openai", "ollama", "gemini", "rest", "universal_rest"];
-    return providersWithEmbedding.includes(provider);
+    return PROVIDERS_WITH_EMBEDDINGS.includes(provider);
   }
   /**
    * Get a human-readable description of the strategy

package/dist/handlers/index.d.mts

@@ -1,3 +1,3 @@
-import '../RagConfig-Crz-l02S.mjs';
-export { c as createChatHandler, a as createHealthHandler, b as createIngestHandler, d as createUploadHandler } from '../index-Cd4CdmEi.mjs';
+import '../RagConfig-Bpp39-um.mjs';
+export { c as createChatHandler, a as createHealthHandler, b as createIngestHandler, d as createUploadHandler } from '../index-CZE72wnV.mjs';
 import 'next/server';

package/dist/handlers/index.d.ts

@@ -1,3 +1,3 @@
-import '../RagConfig-Crz-l02S.js';
-export { c as createChatHandler, a as createHealthHandler, b as createIngestHandler, d as createUploadHandler } from '../index-DQ1pXmlK.js';
+import '../RagConfig-Bpp39-um.js';
+export { c as createChatHandler, a as createHealthHandler, b as createIngestHandler, d as createUploadHandler } from '../index-DSaQwkKA.js';
 import 'next/server';

package/dist/handlers/index.js

@@ -1179,7 +1179,7 @@ var import_server = require("next/server");
 // src/core/ConfigResolver.ts
 init_templateUtils();
 
-// src/config/serverConfig.ts
+// src/config/constants.ts
 var VECTOR_DB_PROVIDERS = [
   "pinecone",
   "pgvector",
@@ -1194,8 +1194,34 @@ var VECTOR_DB_PROVIDERS = [
   "universal_rest",
   "custom"
 ];
-var LLM_PROVIDERS = ["openai", "anthropic", "ollama", "rest", "universal_rest", "custom"];
-var EMBEDDING_PROVIDERS = ["openai", "ollama", "rest", "universal_rest", "custom"];
+var LLM_PROVIDERS = [
+  "openai",
+  "anthropic",
+  "ollama",
+  "gemini",
+  "rest",
+  "universal_rest",
+  "custom"
+];
+var EMBEDDING_PROVIDERS = [
+  "openai",
+  "ollama",
+  "gemini",
+  "rest",
+  "universal_rest",
+  "custom"
+];
+var PROVIDERS_WITH_EMBEDDINGS = [
+  "openai",
+  "ollama",
+  "gemini",
+  "rest",
+  "universal_rest"
+];
+var UI_VISUAL_STYLES = ["glass", "solid"];
+var UI_BORDER_RADIUS_OPTIONS = ["none", "sm", "md", "lg", "xl", "full"];
+
+// src/config/serverConfig.ts
 function readString(env, name) {
   var _a;
   const value = (_a = env[name]) == null ? void 0 : _a.trim();
@@ -1262,6 +1288,7 @@ function getRagConfig(env = process.env) {
   const llmApiKeyByProvider = {
     openai: readString(env, "OPENAI_API_KEY"),
     anthropic: readString(env, "ANTHROPIC_API_KEY"),
+    gemini: readString(env, "GEMINI_API_KEY"),
     ollama: void 0,
     universal_rest: readString(env, "LLM_API_KEY"),
     rest: readString(env, "LLM_API_KEY"),
@@ -1269,6 +1296,7 @@ function getRagConfig(env = process.env) {
   };
   const embeddingApiKeyByProvider = {
     openai: readString(env, "OPENAI_API_KEY"),
+    gemini: readString(env, "GEMINI_API_KEY"),
     ollama: void 0,
     universal_rest: (_q = readString(env, "EMBEDDING_API_KEY")) != null ? _q : readString(env, "OPENAI_API_KEY"),
     custom: (_r = readString(env, "EMBEDDING_API_KEY")) != null ? _r : readString(env, "OPENAI_API_KEY")
@@ -1797,21 +1825,19 @@ var ConfigValidator = class {
       }
     }
     if (config.borderRadius) {
-      const validValues = ["none", "sm", "md", "lg", "xl", "full"];
-      if (!validValues.includes(config.borderRadius)) {
+      if (!UI_BORDER_RADIUS_OPTIONS.includes(config.borderRadius)) {
         errors.push({
           field: "ui.borderRadius",
-          message: `borderRadius must be one of: ${validValues.join(", ")}`,
+          message: `borderRadius must be one of: ${UI_BORDER_RADIUS_OPTIONS.join(", ")}`,
           severity: "warning"
         });
       }
     }
     if (config.visualStyle) {
-      const validValues = ["glass", "solid"];
-      if (!validValues.includes(config.visualStyle)) {
+      if (!UI_VISUAL_STYLES.includes(config.visualStyle)) {
         errors.push({
           field: "ui.visualStyle",
-          message: `visualStyle must be one of: ${validValues.join(", ")}`,
+          message: `visualStyle must be one of: ${UI_VISUAL_STYLES.join(", ")}`,
           severity: "warning"
         });
       }
@@ -2151,8 +2177,18 @@ var GeminiProvider = class {
   constructor(llmConfig, embeddingConfig) {
     if (!llmConfig.apiKey) throw new Error("[GeminiProvider] llmConfig.apiKey is required");
     this.client = new import_genai.GoogleGenAI({ apiKey: llmConfig.apiKey });
-    this.llmConfig = llmConfig;
-    this.embeddingConfig = embeddingConfig;
+    this.llmConfig = __spreadProps(__spreadValues({}, llmConfig), {
+      model: this.sanitizeModel(llmConfig.model)
+    });
+    if (embeddingConfig) {
+      this.embeddingConfig = __spreadProps(__spreadValues({}, embeddingConfig), {
+        model: this.sanitizeModel(embeddingConfig.model)
+      });
+    }
+  }
+  sanitizeModel(model) {
+    if (!model) return model;
+    return model.split(":")[0];
   }
   async chat(messages, context, options) {
     var _a, _b, _c, _d, _e, _f;
@@ -2181,8 +2217,10 @@ ${context}`;
     return (_f = response.text) != null ? _f : "";
   }
   async embed(text, options) {
-    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j;
-    const model = (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004";
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k;
+    const model = this.sanitizeModel(
+      (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004"
+    );
     const apiKey = (_e = (_d = this.embeddingConfig) == null ? void 0 : _d.apiKey) != null ? _e : this.llmConfig.apiKey;
     const client = apiKey !== this.llmConfig.apiKey ? new import_genai.GoogleGenAI({ apiKey }) : this.client;
     let content = text;
@@ -2199,14 +2237,20 @@ ${context}`;
     }
     const response = await client.models.embedContent({
       model,
-      contents: content
+      contents: content,
+      config: ((_h = this.embeddingConfig) == null ? void 0 : _h.dimensions) ? { outputDimensionality: this.embeddingConfig.dimensions } : void 0
+    }).catch((err) => {
+      console.error(`[GeminiProvider] Embedding failed for model "${model}":`, err.message);
+      throw err;
     });
-    return (_j = (_i = (_h = response.embeddings) == null ? void 0 : _h[0]) == null ? void 0 : _i.values) != null ? _j : [];
+    return (_k = (_j = (_i = response.embeddings) == null ? void 0 : _i[0]) == null ? void 0 : _j.values) != null ? _k : [];
   }
   async batchEmbed(texts, options) {
-    var _a, _b, _c, _d, _e, _f, _g, _h, _i;
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j;
     const vectors = [];
-    const model = (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004";
+    const model = this.sanitizeModel(
+      (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004"
+    );
     const apiKey = (_e = (_d = this.embeddingConfig) == null ? void 0 : _d.apiKey) != null ? _e : this.llmConfig.apiKey;
     const client = apiKey !== this.llmConfig.apiKey ? new import_genai.GoogleGenAI({ apiKey }) : this.client;
     let contents = texts;
@@ -2219,12 +2263,16 @@ ${context}`;
     }
     const response = await client.models.embedContent({
       model,
-      contents
+      contents,
+      config: ((_h = this.embeddingConfig) == null ? void 0 : _h.dimensions) ? { outputDimensionality: this.embeddingConfig.dimensions } : void 0
+    }).catch((err) => {
+      console.error(`[GeminiProvider] Batch embedding failed for model "${model}":`, err.message);
+      throw err;
     });
-    return (_i = (_h = response.embeddings) == null ? void 0 : _h.map((e) => {
+    return (_j = (_i = response.embeddings) == null ? void 0 : _i.map((e) => {
       var _a2;
       return (_a2 = e.values) != null ? _a2 : [];
-    })) != null ? _i : [];
+    })) != null ? _j : [];
   }
   async ping() {
     try {
@@ -2785,8 +2833,7 @@ var EmbeddingStrategyResolver = class {
    * Check if an LLM provider natively supports embeddings
    */
   static supportsEmbedding(provider) {
-    const providersWithEmbedding = ["openai", "ollama", "gemini", "rest", "universal_rest"];
-    return providersWithEmbedding.includes(provider);
+    return PROVIDERS_WITH_EMBEDDINGS.includes(provider);
   }
   /**
    * Get a human-readable description of the strategy

package/dist/handlers/index.mjs

@@ -3,7 +3,7 @@ import {
   createHealthHandler,
   createIngestHandler,
   createUploadHandler
-} from "../chunk-7BPQDPOC.mjs";
+} from "../chunk-GE4AVGA5.mjs";
 import "../chunk-EDLTMSNY.mjs";
 import "../chunk-FWCSY2DS.mjs";
 export {

package/dist/{index-Cd4CdmEi.d.mts → index-CZE72wnV.d.mts}

@@ -1,4 +1,4 @@
-import { c as VectorDBConfig, L as LLMConfig, E as EmbeddingConfig, a as RagConfig, C as ChatResponse } from './RagConfig-Crz-l02S.mjs';
+import { c as VectorDBConfig, L as LLMConfig, E as EmbeddingConfig, a as RagConfig, C as ChatResponse } from './RagConfig-Bpp39-um.mjs';
 import { NextRequest, NextResponse } from 'next/server';
 
 /**

package/dist/{index-DQ1pXmlK.d.ts → index-DSaQwkKA.d.ts}

@@ -1,4 +1,4 @@
-import { c as VectorDBConfig, L as LLMConfig, E as EmbeddingConfig, a as RagConfig, C as ChatResponse } from './RagConfig-Crz-l02S.js';
+import { c as VectorDBConfig, L as LLMConfig, E as EmbeddingConfig, a as RagConfig, C as ChatResponse } from './RagConfig-Bpp39-um.js';
 import { NextRequest, NextResponse } from 'next/server';
 
 /**

package/dist/index.d.mts

@@ -2,8 +2,8 @@ import * as react_jsx_runtime from 'react/jsx-runtime';
 import React$1, { ReactNode } from 'react';
 import { C as ChatMessage } from './DocumentChunker-BICIjSuG.mjs';
 export { a as ChatOptions, b as Chunk, c as ChunkOptions, E as EmbedOptions, I as ILLMProvider } from './DocumentChunker-BICIjSuG.mjs';
-import { V as VectorMatch, U as UIConfig } from './RagConfig-Crz-l02S.mjs';
-export { C as ChatResponse, E as EmbeddingConfig, I as IngestDocument, L as LLMConfig, R as RAGConfig, a as RagConfig, b as UpsertDocument, c as VectorDBConfig } from './RagConfig-Crz-l02S.mjs';
+import { V as VectorMatch, U as UIConfig } from './RagConfig-Bpp39-um.mjs';
+export { C as ChatResponse, E as EmbeddingConfig, I as IngestDocument, L as LLMConfig, R as RAGConfig, a as RagConfig, b as UpsertDocument, c as VectorDBConfig } from './RagConfig-Bpp39-um.mjs';
 
 interface ChatWidgetProps {
     /** Position of the floating button. Defaults to bottom-right. */

package/dist/index.d.ts

@@ -2,8 +2,8 @@ import * as react_jsx_runtime from 'react/jsx-runtime';
 import React$1, { ReactNode } from 'react';
 import { C as ChatMessage } from './DocumentChunker-BICIjSuG.js';
 export { a as ChatOptions, b as Chunk, c as ChunkOptions, E as EmbedOptions, I as ILLMProvider } from './DocumentChunker-BICIjSuG.js';
-import { V as VectorMatch, U as UIConfig } from './RagConfig-Crz-l02S.js';
-export { C as ChatResponse, E as EmbeddingConfig, I as IngestDocument, L as LLMConfig, R as RAGConfig, a as RagConfig, b as UpsertDocument, c as VectorDBConfig } from './RagConfig-Crz-l02S.js';
+import { V as VectorMatch, U as UIConfig } from './RagConfig-Bpp39-um.js';
+export { C as ChatResponse, E as EmbeddingConfig, I as IngestDocument, L as LLMConfig, R as RAGConfig, a as RagConfig, b as UpsertDocument, c as VectorDBConfig } from './RagConfig-Bpp39-um.js';
 
 interface ChatWidgetProps {
     /** Position of the floating button. Defaults to bottom-right. */

package/dist/server.d.mts

@@ -1,9 +1,9 @@
-import { a as RagConfig, C as ChatResponse, I as IngestDocument, c as VectorDBConfig, b as UpsertDocument, V as VectorMatch, L as LLMConfig, E as EmbeddingConfig, d as VectorDBProvider, e as LLMProvider, f as EmbeddingProvider } from './RagConfig-Crz-l02S.mjs';
-export { R as RAGConfig, U as UIConfig } from './RagConfig-Crz-l02S.mjs';
+import { a as RagConfig, C as ChatResponse, I as IngestDocument, c as VectorDBConfig, b as UpsertDocument, V as VectorMatch, L as LLMConfig, E as EmbeddingConfig, d as VectorDBProvider, e as LLMProvider, f as EmbeddingProvider } from './RagConfig-Bpp39-um.mjs';
+export { R as RAGConfig, U as UIConfig } from './RagConfig-Bpp39-um.mjs';
 import { C as ChatMessage, I as ILLMProvider, a as ChatOptions, E as EmbedOptions } from './DocumentChunker-BICIjSuG.mjs';
 export { b as Chunk, c as ChunkOptions, D as DocumentChunker } from './DocumentChunker-BICIjSuG.mjs';
-import { H as HealthCheckResult } from './index-Cd4CdmEi.mjs';
-export { P as ProviderHealthCheck, c as createChatHandler, a as createHealthHandler, b as createIngestHandler, d as createUploadHandler } from './index-Cd4CdmEi.mjs';
+import { H as HealthCheckResult } from './index-CZE72wnV.mjs';
+export { P as ProviderHealthCheck, c as createChatHandler, a as createHealthHandler, b as createIngestHandler, d as createUploadHandler } from './index-CZE72wnV.mjs';
 import 'next/server';
 
 /**

package/dist/server.d.ts

@@ -1,9 +1,9 @@
-import { a as RagConfig, C as ChatResponse, I as IngestDocument, c as VectorDBConfig, b as UpsertDocument, V as VectorMatch, L as LLMConfig, E as EmbeddingConfig, d as VectorDBProvider, e as LLMProvider, f as EmbeddingProvider } from './RagConfig-Crz-l02S.js';
-export { R as RAGConfig, U as UIConfig } from './RagConfig-Crz-l02S.js';
+import { a as RagConfig, C as ChatResponse, I as IngestDocument, c as VectorDBConfig, b as UpsertDocument, V as VectorMatch, L as LLMConfig, E as EmbeddingConfig, d as VectorDBProvider, e as LLMProvider, f as EmbeddingProvider } from './RagConfig-Bpp39-um.js';
+export { R as RAGConfig, U as UIConfig } from './RagConfig-Bpp39-um.js';
 import { C as ChatMessage, I as ILLMProvider, a as ChatOptions, E as EmbedOptions } from './DocumentChunker-BICIjSuG.js';
 export { b as Chunk, c as ChunkOptions, D as DocumentChunker } from './DocumentChunker-BICIjSuG.js';
-import { H as HealthCheckResult } from './index-DQ1pXmlK.js';
-export { P as ProviderHealthCheck, c as createChatHandler, a as createHealthHandler, b as createIngestHandler, d as createUploadHandler } from './index-DQ1pXmlK.js';
+import { H as HealthCheckResult } from './index-DSaQwkKA.js';
+export { P as ProviderHealthCheck, c as createChatHandler, a as createHealthHandler, b as createIngestHandler, d as createUploadHandler } from './index-DSaQwkKA.js';
 import 'next/server';
 
 /**

package/dist/server.js

@@ -1223,7 +1223,7 @@ module.exports = __toCommonJS(server_exports);
 // src/core/ConfigResolver.ts
 init_templateUtils();
 
-// src/config/serverConfig.ts
+// src/config/constants.ts
 var VECTOR_DB_PROVIDERS = [
   "pinecone",
   "pgvector",
@@ -1238,8 +1238,34 @@ var VECTOR_DB_PROVIDERS = [
   "universal_rest",
   "custom"
 ];
-var LLM_PROVIDERS = ["openai", "anthropic", "ollama", "rest", "universal_rest", "custom"];
-var EMBEDDING_PROVIDERS = ["openai", "ollama", "rest", "universal_rest", "custom"];
+var LLM_PROVIDERS = [
+  "openai",
+  "anthropic",
+  "ollama",
+  "gemini",
+  "rest",
+  "universal_rest",
+  "custom"
+];
+var EMBEDDING_PROVIDERS = [
+  "openai",
+  "ollama",
+  "gemini",
+  "rest",
+  "universal_rest",
+  "custom"
+];
+var PROVIDERS_WITH_EMBEDDINGS = [
+  "openai",
+  "ollama",
+  "gemini",
+  "rest",
+  "universal_rest"
+];
+var UI_VISUAL_STYLES = ["glass", "solid"];
+var UI_BORDER_RADIUS_OPTIONS = ["none", "sm", "md", "lg", "xl", "full"];
+
+// src/config/serverConfig.ts
 function readString(env, name) {
   var _a;
   const value = (_a = env[name]) == null ? void 0 : _a.trim();
@@ -1306,6 +1332,7 @@ function getRagConfig(env = process.env) {
   const llmApiKeyByProvider = {
     openai: readString(env, "OPENAI_API_KEY"),
     anthropic: readString(env, "ANTHROPIC_API_KEY"),
+    gemini: readString(env, "GEMINI_API_KEY"),
     ollama: void 0,
     universal_rest: readString(env, "LLM_API_KEY"),
     rest: readString(env, "LLM_API_KEY"),
@@ -1313,6 +1340,7 @@ function getRagConfig(env = process.env) {
   };
   const embeddingApiKeyByProvider = {
     openai: readString(env, "OPENAI_API_KEY"),
+    gemini: readString(env, "GEMINI_API_KEY"),
     ollama: void 0,
     universal_rest: (_q = readString(env, "EMBEDDING_API_KEY")) != null ? _q : readString(env, "OPENAI_API_KEY"),
     custom: (_r = readString(env, "EMBEDDING_API_KEY")) != null ? _r : readString(env, "OPENAI_API_KEY")
@@ -1841,21 +1869,19 @@ var ConfigValidator = class {
       }
     }
     if (config.borderRadius) {
-      const validValues = ["none", "sm", "md", "lg", "xl", "full"];
-      if (!validValues.includes(config.borderRadius)) {
+      if (!UI_BORDER_RADIUS_OPTIONS.includes(config.borderRadius)) {
         errors.push({
           field: "ui.borderRadius",
-          message: `borderRadius must be one of: ${validValues.join(", ")}`,
+          message: `borderRadius must be one of: ${UI_BORDER_RADIUS_OPTIONS.join(", ")}`,
           severity: "warning"
         });
       }
     }
     if (config.visualStyle) {
-      const validValues = ["glass", "solid"];
-      if (!validValues.includes(config.visualStyle)) {
+      if (!UI_VISUAL_STYLES.includes(config.visualStyle)) {
         errors.push({
           field: "ui.visualStyle",
-          message: `visualStyle must be one of: ${validValues.join(", ")}`,
+          message: `visualStyle must be one of: ${UI_VISUAL_STYLES.join(", ")}`,
           severity: "warning"
         });
       }
@@ -2195,8 +2221,18 @@ var GeminiProvider = class {
   constructor(llmConfig, embeddingConfig) {
     if (!llmConfig.apiKey) throw new Error("[GeminiProvider] llmConfig.apiKey is required");
     this.client = new import_genai.GoogleGenAI({ apiKey: llmConfig.apiKey });
-    this.llmConfig = llmConfig;
-    this.embeddingConfig = embeddingConfig;
+    this.llmConfig = __spreadProps(__spreadValues({}, llmConfig), {
+      model: this.sanitizeModel(llmConfig.model)
+    });
+    if (embeddingConfig) {
+      this.embeddingConfig = __spreadProps(__spreadValues({}, embeddingConfig), {
+        model: this.sanitizeModel(embeddingConfig.model)
+      });
+    }
+  }
+  sanitizeModel(model) {
+    if (!model) return model;
+    return model.split(":")[0];
   }
   async chat(messages, context, options) {
     var _a, _b, _c, _d, _e, _f;
@@ -2225,8 +2261,10 @@ ${context}`;
     return (_f = response.text) != null ? _f : "";
   }
   async embed(text, options) {
-    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j;
-    const model = (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004";
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k;
+    const model = this.sanitizeModel(
+      (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004"
+    );
     const apiKey = (_e = (_d = this.embeddingConfig) == null ? void 0 : _d.apiKey) != null ? _e : this.llmConfig.apiKey;
     const client = apiKey !== this.llmConfig.apiKey ? new import_genai.GoogleGenAI({ apiKey }) : this.client;
     let content = text;
@@ -2243,14 +2281,20 @@ ${context}`;
     }
     const response = await client.models.embedContent({
       model,
-      contents: content
+      contents: content,
+      config: ((_h = this.embeddingConfig) == null ? void 0 : _h.dimensions) ? { outputDimensionality: this.embeddingConfig.dimensions } : void 0
+    }).catch((err) => {
+      console.error(`[GeminiProvider] Embedding failed for model "${model}":`, err.message);
+      throw err;
     });
-    return (_j = (_i = (_h = response.embeddings) == null ? void 0 : _h[0]) == null ? void 0 : _i.values) != null ? _j : [];
+    return (_k = (_j = (_i = response.embeddings) == null ? void 0 : _i[0]) == null ? void 0 : _j.values) != null ? _k : [];
   }
   async batchEmbed(texts, options) {
-    var _a, _b, _c, _d, _e, _f, _g, _h, _i;
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j;
     const vectors = [];
-    const model = (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004";
+    const model = this.sanitizeModel(
+      (_c = (_b = options == null ? void 0 : options.model) != null ? _b : (_a = this.embeddingConfig) == null ? void 0 : _a.model) != null ? _c : "text-embedding-004"
+    );
     const apiKey = (_e = (_d = this.embeddingConfig) == null ? void 0 : _d.apiKey) != null ? _e : this.llmConfig.apiKey;
     const client = apiKey !== this.llmConfig.apiKey ? new import_genai.GoogleGenAI({ apiKey }) : this.client;
     let contents = texts;
@@ -2263,12 +2307,16 @@ ${context}`;
     }
     const response = await client.models.embedContent({
       model,
-      contents
+      contents,
+      config: ((_h = this.embeddingConfig) == null ? void 0 : _h.dimensions) ? { outputDimensionality: this.embeddingConfig.dimensions } : void 0
+    }).catch((err) => {
+      console.error(`[GeminiProvider] Batch embedding failed for model "${model}":`, err.message);
+      throw err;
     });
-    return (_i = (_h = response.embeddings) == null ? void 0 : _h.map((e) => {
+    return (_j = (_i = response.embeddings) == null ? void 0 : _i.map((e) => {
       var _a2;
       return (_a2 = e.values) != null ? _a2 : [];
-    })) != null ? _i : [];
+    })) != null ? _j : [];
   }
   async ping() {
     try {
@@ -2874,8 +2922,7 @@ var EmbeddingStrategyResolver = class {
    * Check if an LLM provider natively supports embeddings
    */
   static supportsEmbedding(provider) {
-    const providersWithEmbedding = ["openai", "ollama", "gemini", "rest", "universal_rest"];
-    return providersWithEmbedding.includes(provider);
+    return PROVIDERS_WITH_EMBEDDINGS.includes(provider);
   }
   /**
    * Get a human-readable description of the strategy

package/dist/server.mjs

@@ -34,7 +34,7 @@ import {
   createIngestHandler,
   createUploadHandler,
   getRagConfig
-} from "./chunk-7BPQDPOC.mjs";
+} from "./chunk-GE4AVGA5.mjs";
 import "./chunk-EDLTMSNY.mjs";
 import {
   PineconeProvider

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@retrivora-ai/rag-engine",
-  "version": "0.3.2",
+  "version": "0.3.4",
   "description": "Retrivora AI is a plug-and-play AI engine for RAG chat experiences — generic vector DB + LLM provider, embeddable or standalone.",
   "author": "Abhinav Alkuchi",
   "license": "MIT",

package/src/config/EmbeddingStrategy.ts

@@ -12,6 +12,7 @@
 import { ILLMProvider } from '../llm/ILLMProvider';
 import { LLMFactory } from '../llm/LLMFactory';
 import { LLMConfig, EmbeddingConfig, LLMProvider, EmbeddingProvider } from './RagConfig';
+import { PROVIDERS_WITH_EMBEDDINGS } from './constants';
 
 export enum EmbeddingStrategy {
   /**
@@ -125,8 +126,7 @@ export class EmbeddingStrategyResolver {
    * Check if an LLM provider natively supports embeddings
    */
   private static supportsEmbedding(provider: LLMProvider | string): boolean {
-    const providersWithEmbedding = ['openai', 'ollama', 'gemini', 'rest', 'universal_rest'];
-    return providersWithEmbedding.includes(provider);
+    return (PROVIDERS_WITH_EMBEDDINGS as readonly string[]).includes(provider);
   }
 
   /**

package/src/config/RagConfig.ts

@@ -8,7 +8,13 @@
 // Vector DB
 // ---------------------------------------------------------------------------
 
-export type VectorDBProvider = 'pinecone' | 'pgvector' | 'postgresql' | 'mongodb' | 'milvus' | 'qdrant' | 'chromadb' | 'redis' | 'weaviate' | 'rest' | 'universal_rest' | 'custom';
+import {
+  VECTOR_DB_PROVIDERS,
+  LLM_PROVIDERS,
+  EMBEDDING_PROVIDERS,
+} from './constants';
+
+export type VectorDBProvider = typeof VECTOR_DB_PROVIDERS[number];
 
 export interface VectorDBConfig {
   /** Which vector database to use */
@@ -40,7 +46,7 @@ export interface VectorDBConfig {
 // LLM Provider
 // ---------------------------------------------------------------------------
 
-export type LLMProvider = 'openai' | 'anthropic' | 'ollama' | 'gemini' | 'rest' | 'universal_rest' | 'custom';
+export type LLMProvider = typeof LLM_PROVIDERS[number];
 
 export interface LLMConfig {
   /** Which LLM provider to use */
@@ -72,7 +78,7 @@ export interface LLMConfig {
 // Embedding
 // ---------------------------------------------------------------------------
 
-export type EmbeddingProvider = 'openai' | 'ollama' | 'gemini' | 'rest' | 'universal_rest' | 'custom';
+export type EmbeddingProvider = typeof EMBEDDING_PROVIDERS[number];
 
 export interface EmbeddingConfig {
   /** Which embedding provider to use */

package/src/config/constants.ts

@@ -0,0 +1,52 @@
+/**
+ * constants.ts — Centralized definitions of acceptable configuration values.
+ * 
+ * This file serves as the single source of truth for all statically supported
+ * providers and stylistic options in the Retrivora AI RAG Engine.
+ */
+
+export const VECTOR_DB_PROVIDERS = [
+  'pinecone',
+  'pgvector',
+  'postgresql',
+  'mongodb',
+  'milvus',
+  'qdrant',
+  'chromadb',
+  'redis',
+  'weaviate',
+  'rest',
+  'universal_rest',
+  'custom',
+] as const;
+
+export const LLM_PROVIDERS = [
+  'openai',
+  'anthropic',
+  'ollama',
+  'gemini',
+  'rest',
+  'universal_rest',
+  'custom',
+] as const;
+
+export const EMBEDDING_PROVIDERS = [
+  'openai',
+  'ollama',
+  'gemini',
+  'rest',
+  'universal_rest',
+  'custom',
+] as const;
+
+export const PROVIDERS_WITH_EMBEDDINGS = [
+  'openai',
+  'ollama',
+  'gemini',
+  'rest',
+  'universal_rest',
+] as const;
+
+export const UI_VISUAL_STYLES = ['glass', 'solid'] as const;
+
+export const UI_BORDER_RADIUS_OPTIONS = ['none', 'sm', 'md', 'lg', 'xl', 'full'] as const;

package/src/config/serverConfig.ts

@@ -19,14 +19,11 @@
  */
 
 import { RagConfig } from './RagConfig';
-
-const VECTOR_DB_PROVIDERS = [
-  'pinecone', 'pgvector', 'postgresql', 'mongodb', 'milvus',
-  'qdrant', 'chromadb', 'redis', 'weaviate', 'rest', 'universal_rest', 'custom',
-] as const;
-
-const LLM_PROVIDERS = ['openai', 'anthropic', 'ollama', 'rest', 'universal_rest', 'custom'] as const;
-const EMBEDDING_PROVIDERS = ['openai', 'ollama', 'rest', 'universal_rest', 'custom'] as const;
+import {
+  VECTOR_DB_PROVIDERS,
+  LLM_PROVIDERS,
+  EMBEDDING_PROVIDERS,
+} from './constants';
 
 function readString(env: Record<string, string | undefined>, name: string): string | undefined {
   const value = env[name]?.trim();
@@ -112,6 +109,7 @@ export function getRagConfig(env: Record<string, string | undefined> = process.e
   const llmApiKeyByProvider: Record<string, string | undefined> = {
     openai: readString(env, 'OPENAI_API_KEY'),
     anthropic: readString(env, 'ANTHROPIC_API_KEY'),
+    gemini: readString(env, 'GEMINI_API_KEY'),
     ollama: undefined,
     universal_rest: readString(env, 'LLM_API_KEY'),
     rest: readString(env, 'LLM_API_KEY'),
@@ -120,6 +118,7 @@ export function getRagConfig(env: Record<string, string | undefined> = process.e
 
   const embeddingApiKeyByProvider: Record<string, string | undefined> = {
     openai: readString(env, 'OPENAI_API_KEY'),
+    gemini: readString(env, 'GEMINI_API_KEY'),
     ollama: undefined,
     universal_rest: readString(env, 'EMBEDDING_API_KEY') ?? readString(env, 'OPENAI_API_KEY'),
     custom: readString(env, 'EMBEDDING_API_KEY') ?? readString(env, 'OPENAI_API_KEY'),

package/src/core/ConfigValidator.ts

@@ -10,6 +10,10 @@
  */
 
 import { RagConfig, VectorDBConfig, LLMConfig, EmbeddingConfig } from '../config/RagConfig';
+import {
+  UI_VISUAL_STYLES,
+  UI_BORDER_RADIUS_OPTIONS,
+} from '../config/constants';
 
 export interface ValidationError {
   field: string;
@@ -508,22 +512,20 @@ export class ConfigValidator {
     }
 
     if (config.borderRadius) {
-      const validValues = ['none', 'sm', 'md', 'lg', 'xl', 'full'];
-      if (!validValues.includes(config.borderRadius as string)) {
+      if (!UI_BORDER_RADIUS_OPTIONS.includes(config.borderRadius as any)) {
         errors.push({
           field: 'ui.borderRadius',
-          message: `borderRadius must be one of: ${validValues.join(', ')}`,
+          message: `borderRadius must be one of: ${UI_BORDER_RADIUS_OPTIONS.join(', ')}`,
           severity: 'warning',
         });
       }
     }
 
     if (config.visualStyle) {
-      const validValues = ['glass', 'solid'];
-      if (!validValues.includes(config.visualStyle as string)) {
+      if (!UI_VISUAL_STYLES.includes(config.visualStyle as any)) {
         errors.push({
           field: 'ui.visualStyle',
-          message: `visualStyle must be one of: ${validValues.join(', ')}`,
+          message: `visualStyle must be one of: ${UI_VISUAL_STYLES.join(', ')}`,
           severity: 'warning',
         });
       }

package/src/llm/providers/GeminiProvider.ts

@@ -26,10 +26,28 @@ export class GeminiProvider implements ILLMProvider {
   constructor(llmConfig: LLMConfig, embeddingConfig?: EmbeddingConfig) {
     if (!llmConfig.apiKey) throw new Error('[GeminiProvider] llmConfig.apiKey is required');
     this.client = new GoogleGenAI({ apiKey: llmConfig.apiKey });
-    this.llmConfig = llmConfig;
-    this.embeddingConfig = embeddingConfig;
+    
+    // Sanitize model names (e.g. remove ':latest' suffix used by Ollama)
+    this.llmConfig = {
+      ...llmConfig,
+      model: this.sanitizeModel(llmConfig.model)
+    };
+    
+    if (embeddingConfig) {
+      this.embeddingConfig = {
+        ...embeddingConfig,
+        model: this.sanitizeModel(embeddingConfig.model)
+      };
+    }
   }
 
+  private sanitizeModel(model: string): string {
+    if (!model) return model;
+    // Strip :latest suffix common in Ollama but invalid in Google SDK
+    return model.split(':')[0];
+  }
+
+
   async chat(messages: ChatMessage[], context: string, options?: ChatOptions): Promise<string> {
     const systemPrompt =
       this.llmConfig.systemPrompt ??
@@ -62,10 +80,11 @@ export class GeminiProvider implements ILLMProvider {
   }
 
   async embed(text: string, options?: EmbedOptions): Promise<number[]> {
-    const model =
+    const model = this.sanitizeModel(
       options?.model ??
       this.embeddingConfig?.model ??
-      'text-embedding-004';
+      'text-embedding-004'
+    );
 
     const apiKey = this.embeddingConfig?.apiKey ?? this.llmConfig.apiKey;
     const client = apiKey !== this.llmConfig.apiKey
@@ -91,6 +110,10 @@ export class GeminiProvider implements ILLMProvider {
     const response = await client.models.embedContent({
       model,
       contents: content,
+      config: this.embeddingConfig?.dimensions ? { outputDimensionality: this.embeddingConfig.dimensions } : undefined
+    }).catch(err => {
+      console.error(`[GeminiProvider] Embedding failed for model "${model}":`, err.message);
+      throw err;
     });
 
     return response.embeddings?.[0]?.values ?? [];
@@ -101,10 +124,11 @@ export class GeminiProvider implements ILLMProvider {
     // Sequential fallback for Gemini (API doesn't have a direct batch endpoint in the same way, though you can pass multiple contents.
     // For simplicity and to match the other providers' structure, we'll iterate or pass array.
     // The new SDK supports an array of contents.
-    const model =
+    const model = this.sanitizeModel(
       options?.model ??
       this.embeddingConfig?.model ??
-      'text-embedding-004';
+      'text-embedding-004'
+    );
 
     const apiKey = this.embeddingConfig?.apiKey ?? this.llmConfig.apiKey;
     const client = apiKey !== this.llmConfig.apiKey
@@ -126,6 +150,10 @@ export class GeminiProvider implements ILLMProvider {
     const response = await client.models.embedContent({
       model,
       contents,
+      config: this.embeddingConfig?.dimensions ? { outputDimensionality: this.embeddingConfig.dimensions } : undefined
+    }).catch(err => {
+      console.error(`[GeminiProvider] Batch embedding failed for model "${model}":`, err.message);
+      throw err;
     });
 
     return response.embeddings?.map(e => e.values ?? []) ?? [];