feat(knowledge): support Voyage AI (#3810)

* feat(knowledge): support Voyage AI

* chore

package.json

@@ -53,6 +53,7 @@
     "@electron-toolkit/utils": "^3.0.0",
     "@electron/notarize": "^2.5.0",
     "@google/generative-ai": "^0.21.0",
+    "@langchain/community": "^0.3.36",
     "@llm-tools/embedjs": "patch:@llm-tools/embedjs@npm%3A0.1.28#~/.yarn/patches/@llm-tools-embedjs-npm-0.1.28-8e4393fa2d.patch",
     "@llm-tools/embedjs-libsql": "^0.1.28",
     "@llm-tools/embedjs-loader-csv": "^0.1.28",

src/main/embeddings/Embeddings.ts

@@ -0,0 +1,24 @@
+import type { BaseEmbeddings } from '@llm-tools/embedjs-interfaces'
+import { KnowledgeBaseParams } from '@types'
+
+import EmbeddingsFactory from './EmbeddingsFactory'
+
+export default class Embeddings {
+  private sdk: BaseEmbeddings
+  constructor({ model, apiKey, apiVersion, baseURL, dimensions }: KnowledgeBaseParams) {
+    this.sdk = EmbeddingsFactory.create({ model, apiKey, apiVersion, baseURL, dimensions } as KnowledgeBaseParams)
+  }
+  public async init(): Promise<void> {
+    return this.sdk.init()
+  }
+  public async getDimensions(): Promise<number> {
+    return this.sdk.getDimensions()
+  }
+  public async embedDocuments(texts: string[]): Promise<number[][]> {
+    return this.sdk.embedDocuments(texts)
+  }
+
+  public async embedQuery(text: string): Promise<number[]> {
+    return this.sdk.embedQuery(text)
+  }
+}

src/main/embeddings/EmbeddingsFactory.ts

@@ -0,0 +1,38 @@
+import type { BaseEmbeddings } from '@llm-tools/embedjs-interfaces'
+import { OpenAiEmbeddings } from '@llm-tools/embedjs-openai'
+import { AzureOpenAiEmbeddings } from '@llm-tools/embedjs-openai/src/azure-openai-embeddings'
+import { getInstanceName } from '@main/utils'
+import { KnowledgeBaseParams } from '@types'
+
+import VoyageEmbeddings from './VoyageEmbeddings'
+
+export default class EmbeddingsFactory {
+  static create({ model, apiKey, apiVersion, baseURL, dimensions }: KnowledgeBaseParams): BaseEmbeddings {
+    const batchSize = 10
+    if (model.includes('voyage')) {
+      return new VoyageEmbeddings({
+        modelName: model,
+        apiKey,
+        outputDimension: dimensions,
+        batchSize: 8
+      })
+    }
+    if (apiVersion !== undefined) {
+      return new AzureOpenAiEmbeddings({
+        azureOpenAIApiKey: apiKey,
+        azureOpenAIApiVersion: apiVersion,
+        azureOpenAIApiDeploymentName: model,
+        azureOpenAIApiInstanceName: getInstanceName(baseURL),
+        dimensions,
+        batchSize
+      })
+    }
+    return new OpenAiEmbeddings({
+      model,
+      apiKey,
+      dimensions,
+      batchSize,
+      configuration: { baseURL }
+    })
+  }
+}

src/main/embeddings/VoyageEmbeddings.ts

@@ -0,0 +1,30 @@
+import { VoyageEmbeddings as _VoyageEmbeddings } from '@langchain/community/embeddings/voyage'
+import { BaseEmbeddings } from '@llm-tools/embedjs-interfaces'
+
+export default class VoyageEmbeddings extends BaseEmbeddings {
+  private model: _VoyageEmbeddings
+  constructor(private readonly configuration?: ConstructorParameters<typeof _VoyageEmbeddings>[0]) {
+    super()
+    if (!this.configuration) this.configuration = {}
+    if (!this.configuration.modelName) this.configuration.modelName = 'voyage-3'
+
+    if (!this.configuration.outputDimension) {
+      throw new Error('You need to pass in the optional dimensions parameter for this model')
+    }
+    this.model = new _VoyageEmbeddings(this.configuration)
+  }
+  override async getDimensions(): Promise<number> {
+    if (!this.configuration?.outputDimension) {
+      throw new Error('You need to pass in the optional dimensions parameter for this model')
+    }
+    return this.configuration?.outputDimension
+  }
+
+  override async embedDocuments(texts: string[]): Promise<number[][]> {
+    return this.model.embedDocuments(texts)
+  }
+
+  override async embedQuery(text: string): Promise<number[]> {
+    return this.model.embedQuery(text)
+  }
+}

src/main/reranker/RerankerFactory.ts

@@ -4,6 +4,7 @@ import BaseReranker from './BaseReranker'
 import DefaultReranker from './DefaultReranker'
 import JinaReranker from './JinaReranker'
 import SiliconFlowReranker from './SiliconFlowReranker'
+import VoyageReranker from './VoyageReranker'
 
 export default class RerankerFactory {
   static create(base: KnowledgeBaseParams): BaseReranker {
@@ -11,6 +12,8 @@ export default class RerankerFactory {
       return new SiliconFlowReranker(base)
     } else if (base.rerankModelProvider === 'jina') {
       return new JinaReranker(base)
+    } else if (base.rerankModelProvider === 'voyageai') {
+      return new VoyageReranker(base)
     }
     return new DefaultReranker(base)
   }

src/main/reranker/VoyageReranker.ts

@@ -0,0 +1,60 @@
+import { ExtractChunkData } from '@llm-tools/embedjs-interfaces'
+import { KnowledgeBaseParams } from '@types'
+import axios from 'axios'
+
+import BaseReranker from './BaseReranker'
+
+export default class VoyageReranker extends BaseReranker {
+  constructor(base: KnowledgeBaseParams) {
+    super(base)
+  }
+
+  public rerank = async (query: string, searchResults: ExtractChunkData[]): Promise<ExtractChunkData[]> => {
+    let baseURL = this.base?.rerankBaseURL?.endsWith('/')
+      ? this.base.rerankBaseURL.slice(0, -1)
+      : this.base.rerankBaseURL
+
+    if (baseURL && !baseURL.endsWith('/v1')) {
+      baseURL = `${baseURL}/v1`
+    }
+
+    const url = `${baseURL}/rerank`
+
+    const requestBody = {
+      model: this.base.rerankModel,
+      query,
+      documents: searchResults.map((doc) => doc.pageContent),
+      top_k: this.base.topN,
+      return_documents: false,
+      truncation: true
+    }
+
+    try {
+      const { data } = await axios.post(url, requestBody, {
+        headers: {
+          ...this.defaultHeaders()
+        }
+      })
+
+      const rerankResults = data.data
+
+      const resultMap = new Map(rerankResults.map((result: any) => [result.index, result.relevance_score || 0]))
+
+      return searchResults
+        .map((doc: ExtractChunkData, index: number) => {
+          const score = resultMap.get(index)
+          if (score === undefined) return undefined
+
+          return {
+            ...doc,
+            score
+          }
+        })
+        .filter((doc): doc is ExtractChunkData => doc !== undefined)
+        .sort((a, b) => b.score - a.score)
+    } catch (error: any) {
+      console.error('Voyage Reranker API 错误:', error.message || error)
+      throw new Error(`${error} - BaseUrl: ${baseURL}`)
+    }
+  }
+}

src/main/services/KnowledgeService.ts

@@ -21,11 +21,10 @@ import type { ExtractChunkData } from '@llm-tools/embedjs-interfaces'
 import { LibSqlDb } from '@llm-tools/embedjs-libsql'
 import { SitemapLoader } from '@llm-tools/embedjs-loader-sitemap'
 import { WebLoader } from '@llm-tools/embedjs-loader-web'
-import { AzureOpenAiEmbeddings, OpenAiEmbeddings } from '@llm-tools/embedjs-openai'
+import Embeddings from '@main/embeddings/Embeddings'
 import { addFileLoader } from '@main/loader'
 import Reranker from '@main/reranker/Reranker'
 import { windowService } from '@main/services/WindowService'
-import { getInstanceName } from '@main/utils'
 import { getAllFiles } from '@main/utils/file'
 import type { LoaderReturn } from '@shared/config/types'
 import { FileType, KnowledgeBaseParams, KnowledgeItem } from '@types'
@@ -114,29 +113,20 @@ class KnowledgeService {
     baseURL,
     dimensions
   }: KnowledgeBaseParams): Promise<RAGApplication> => {
-    const batchSize = 10
+    let ragApplication: RAGApplication
-    return new RAGApplicationBuilder()
+    const embeddings = new Embeddings({ model, apiKey, apiVersion, baseURL, dimensions } as KnowledgeBaseParams)
-      .setModel('NO_MODEL')
+    try {
-      .setEmbeddingModel(
+      ragApplication = await new RAGApplicationBuilder()
-        apiVersion
+        .setModel('NO_MODEL')
-          ? new AzureOpenAiEmbeddings({
+        .setEmbeddingModel(embeddings)
-              azureOpenAIApiKey: apiKey,
+        .setVectorDatabase(new LibSqlDb({ path: path.join(this.storageDir, id) }))
-              azureOpenAIApiVersion: apiVersion,
+        .build()
-              azureOpenAIApiDeploymentName: model,
+    } catch (e) {
-              azureOpenAIApiInstanceName: getInstanceName(baseURL),
+      Logger.error(e)
-              dimensions,
+      throw new Error(`Failed to create RAGApplication: ${e}`)
-              batchSize
+    }
-            })
+
-          : new OpenAiEmbeddings({
+    return ragApplication
-              model,
-              apiKey,
-              dimensions,
-              batchSize,
-              configuration: { baseURL }
-            })
-      )
-      .setVectorDatabase(new LibSqlDb({ path: path.join(this.storageDir, id) }))
-      .build()
   }
 
   public create = async (_: Electron.IpcMainInvokeEvent, base: KnowledgeBaseParams): Promise<void> => {

src/renderer/src/config/embedings.ts

@@ -242,6 +242,58 @@ export const EMBEDDING_MODELS = [
   {
     id: 'mistral-embed',
     max_context: 8000
+  },
+  {
+    id: 'voyage-3-large',
+    max_context: 1024
+  },
+  {
+    id: 'voyage-3-large',
+    max_context: 256
+  },
+  {
+    id: 'voyage-3-large',
+    max_context: 512
+  },
+  {
+    id: 'voyage-3-large',
+    max_context: 2048
+  },
+  {
+    id: 'voyage-3',
+    max_context: 1024
+  },
+  {
+    id: 'voyage-3-lite',
+    max_context: 512
+  },
+  {
+    id: 'voyage-code-3',
+    max_context: 1024
+  },
+  {
+    id: 'voyage-code-3',
+    max_context: 256
+  },
+  {
+    id: 'voyage-code-3',
+    max_context: 512
+  },
+  {
+    id: 'voyage-code-3',
+    max_context: 2048
+  },
+  {
+    id: 'voyage-finance-2',
+    max_context: 1024
+  },
+  {
+    id: 'voyage-law-2',
+    max_context: 1024
+  },
+  {
+    id: 'voyage-code-2',
+    max_context: 1536
   }
 ]
 

src/renderer/src/config/models.ts

@@ -122,6 +122,7 @@ import UpstageModelLogo from '@renderer/assets/images/models/upstage.png'
 import UpstageModelLogoDark from '@renderer/assets/images/models/upstage_dark.png'
 import ViduModelLogo from '@renderer/assets/images/models/vidu.png'
 import ViduModelLogoDark from '@renderer/assets/images/models/vidu_dark.png'
+import VoyageModelLogo from '@renderer/assets/images/models/voyageai.png'
 import WenxinModelLogo from '@renderer/assets/images/models/wenxin.png'
 import WenxinModelLogoDark from '@renderer/assets/images/models/wenxin_dark.png'
 import XirangModelLogo from '@renderer/assets/images/models/xirang.png'
@@ -175,7 +176,8 @@ export const REASONING_REGEX =
   /^(o\d+(?:-[\w-]+)?|.*\b(?:reasoner|thinking)\b.*|.*-[rR]\d+.*|.*\bqwq(?:-[\w-]+)?\b.*|.*\bhunyuan-t1(?:-[\w-]+)?\b.*)$/i
 
 // Embedding models
-export const EMBEDDING_REGEX = /(?:^text-|embed|bge-|e5-|LLM2Vec|retrieval|uae-|gte-|jina-clip|jina-embeddings)/i
+export const EMBEDDING_REGEX =
+  /(?:^text-|embed|bge-|e5-|LLM2Vec|retrieval|uae-|gte-|jina-clip|jina-embeddings|voyage-)/i
 
 // Rerank models
 export const RERANKING_REGEX = /(?:rerank|re-rank|re-ranker|re-ranking|retrieval|retriever)/i
@@ -327,7 +329,8 @@ export function getModelLogo(modelId: string) {
     embedding: isLight ? EmbeddingModelLogo : EmbeddingModelLogoDark,
     perplexity: isLight ? PerplexityModelLogo : PerplexityModelLogoDark,
     sonar: isLight ? PerplexityModelLogo : PerplexityModelLogoDark,
-    'bge-': BgeModelLogo
+    'bge-': BgeModelLogo,
+    'voyage-': VoyageModelLogo
   }
 
   for (const key in logoMap) {
@@ -1801,7 +1804,63 @@ export const SYSTEM_MODELS: Record<string, Model[]> = {
       group: 'DeepSeek'
     }
   ],
-  gpustack: []
+  gpustack: [],
+  voyageai: [
+    {
+      id: 'voyage-3-large',
+      provider: 'voyageai',
+      name: 'voyage-3-large',
+      group: 'Voyage Embeddings V3'
+    },
+    {
+      id: 'voyage-3',
+      provider: 'voyageai',
+      name: 'voyage-3',
+      group: 'Voyage Embeddings V3'
+    },
+    {
+      id: 'voyage-3-lite',
+      provider: 'voyageai',
+      name: 'voyage-3-lite',
+      group: 'Voyage Embeddings V3'
+    },
+    {
+      id: 'voyage-code-3',
+      provider: 'voyageai',
+      name: 'voyage-code-3',
+      group: 'Voyage Embeddings V3'
+    },
+    {
+      id: 'voyage-finance-3',
+      provider: 'voyageai',
+      name: 'voyage-finance-3',
+      group: 'Voyage Embeddings V2'
+    },
+    {
+      id: 'voyage-law-2',
+      provider: 'voyageai',
+      name: 'voyage-law-2',
+      group: 'Voyage Embeddings V2'
+    },
+    {
+      id: 'voyage-code-2',
+      provider: 'voyageai',
+      name: 'voyage-code-2',
+      group: 'Voyage Embeddings V2'
+    },
+    {
+      id: 'rerank-2',
+      provider: 'voyageai',
+      name: 'rerank-2',
+      group: 'Voyage Rerank V2'
+    },
+    {
+      id: 'rerank-2-lite',
+      provider: 'voyageai',
+      name: 'rerank-2-lite',
+      group: 'Voyage Rerank V2'
+    }
+  ]
 }
 
 export const TEXT_TO_IMAGES_MODELS = [

src/renderer/src/config/providers.ts

@@ -38,6 +38,7 @@ import StepProviderLogo from '@renderer/assets/images/providers/step.png'
 import TencentCloudProviderLogo from '@renderer/assets/images/providers/tencent-cloud-ti.png'
 import TogetherProviderLogo from '@renderer/assets/images/providers/together.png'
 import BytedanceProviderLogo from '@renderer/assets/images/providers/volcengine.png'
+import VoyageAIProviderLogo from '@renderer/assets/images/providers/voyageai.png'
 import XirangProviderLogo from '@renderer/assets/images/providers/xirang.png'
 import ZeroOneProviderLogo from '@renderer/assets/images/providers/zero-one.png'
 import ZhipuProviderLogo from '@renderer/assets/images/providers/zhipu.png'
@@ -86,14 +87,15 @@ const PROVIDER_LOGO_MAP = {
   o3: O3ProviderLogo,
   'tencent-cloud-ti': TencentCloudProviderLogo,
   gpustack: GPUStackProviderLogo,
-  alayanew: AlayaNewProviderLogo
+  alayanew: AlayaNewProviderLogo,
+  voyageai: VoyageAIProviderLogo
 } as const
 
 export function getProviderLogo(providerId: string) {
   return PROVIDER_LOGO_MAP[providerId as keyof typeof PROVIDER_LOGO_MAP]
 }
 
-export const SUPPORTED_REANK_PROVIDERS = ['silicon', 'jina']
+export const SUPPORTED_REANK_PROVIDERS = ['silicon', 'jina', 'voyageai']
 
 export const PROVIDER_CONFIG = {
   openai: {
@@ -560,5 +562,16 @@ export const PROVIDER_CONFIG = {
       docs: 'https://docs.gpustack.ai/latest/',
       models: 'https://docs.gpustack.ai/latest/overview/#supported-models'
     }
+  },
+  voyageai: {
+    api: {
+      url: 'https://api.voyageai.com'
+    },
+    websites: {
+      official: 'https://www.voyageai.com/',
+      apiKey: 'https://dashboard.voyageai.com/organization/api-keys',
+      docs: 'https://docs.voyageai.com/docs',
+      models: 'https://docs.voyageai.com/docs'
+    }
   }
 }

src/renderer/src/i18n/locales/en-us.json

@@ -678,7 +678,8 @@
       "xirang": "State Cloud Xirang",
       "yi": "Yi",
       "zhinao": "360AI",
-      "zhipu": "ZHIPU AI"
+      "zhipu": "ZHIPU AI",
+      "voyageai": "Voyage AI"
     },
     "restore": {
       "confirm": "Are you sure you want to restore data?",

src/renderer/src/i18n/locales/ja-jp.json

@@ -678,7 +678,8 @@
       "xirang": "天翼クラウド 息壤",
       "yi": "零一万物",
       "zhinao": "360智脳",
-      "zhipu": "智譜AI"
+      "zhipu": "智譜AI",
+      "voyageai": "Voyage AI"
     },
     "restore": {
       "confirm": "データを復元しますか？",

src/renderer/src/i18n/locales/ru-ru.json

@@ -678,7 +678,8 @@
       "xirang": "State Cloud Xirang",
       "yi": "Yi",
       "zhinao": "360AI",
-      "zhipu": "ZHIPU AI"
+      "zhipu": "ZHIPU AI",
+      "voyageai": "Voyage AI"
     },
     "restore": {
       "confirm": "Вы уверены, что хотите восстановить данные?",

src/renderer/src/i18n/locales/zh-cn.json

@@ -678,7 +678,8 @@
       "xirang": "天翼云息壤",
       "yi": "零一万物",
       "zhinao": "360智脑",
-      "zhipu": "智谱AI"
+      "zhipu": "智谱AI",
+      "voyageai":"Voyage AI"
     },
     "restore": {
       "confirm": "确定要恢复数据吗？",

src/renderer/src/i18n/locales/zh-tw.json

@@ -678,7 +678,8 @@
       "xirang": "天翼雲息壤",
       "yi": "零一萬物",
       "zhinao": "360 智腦",
-      "zhipu": "智譜 AI"
+      "zhipu": "智譜 AI",
+      "voyageai": "Voyage AI"
     },
     "restore": {
       "confirm": "確定要復原資料嗎？",

src/renderer/src/pages/settings/ProviderSettings/ModelList.tsx

@@ -257,6 +257,7 @@ const ModelList: React.FC<ModelListProps> = ({ provider: _provider, modelStatuse
           {sortedModelGroups[group].map((model) => {
             const modelStatus = modelStatuses.find((status) => status.model.id === model.id)
             const isChecking = modelStatus?.checking === true
+            console.log('model', model.id, getModelLogo(model.id))
 
             return (
               <ModelListItem key={model.id}>

src/renderer/src/store/index.ts

@@ -40,7 +40,7 @@ const persistedReducer = persistReducer(
   {
     key: 'cherry-studio',
     storage,
-    version: 83,
+    version: 84,
     blacklist: ['runtime', 'messages'],
     migrate
   },

src/renderer/src/store/llm.ts

@@ -456,6 +456,16 @@ export const INITIAL_PROVIDERS: Provider[] = [
     models: SYSTEM_MODELS.gpustack,
     isSystem: true,
     enabled: false
+  },
+  {
+    id: 'voyageai',
+    name: 'VoyageAI',
+    type: 'openai',
+    apiKey: '',
+    apiHost: 'https://api.voyageai.com',
+    models: SYSTEM_MODELS.voyageai,
+    isSystem: true,
+    enabled: false
   }
 ]
 

src/renderer/src/store/migrate.ts

@@ -795,6 +795,10 @@ const migrateConfig = {
     state.settings.launchToTray = false
     state.settings.trayOnClose = true
     return state
+  },
+  '84': (state: RootState) => {
+    addProvider(state, 'voyageai')
+    return state
   }
 }
 

yarn.lock

@@ -1590,6 +1590,396 @@ __metadata:
   languageName: node
   linkType: hard
 
+"@langchain/community@npm:^0.3.36":
+  version: 0.3.36
+  resolution: "@langchain/community@npm:0.3.36"
+  dependencies:
+    "@langchain/openai": "npm:>=0.2.0 <0.5.0"
+    binary-extensions: "npm:^2.2.0"
+    expr-eval: "npm:^2.0.2"
+    flat: "npm:^5.0.2"
+    js-yaml: "npm:^4.1.0"
+    langchain: "npm:>=0.2.3 <0.3.0 || >=0.3.4 <0.4.0"
+    langsmith: "npm:>=0.2.8 <0.4.0"
+    uuid: "npm:^10.0.0"
+    zod: "npm:^3.22.3"
+    zod-to-json-schema: "npm:^3.22.5"
+  peerDependencies:
+    "@arcjet/redact": ^v1.0.0-alpha.23
+    "@aws-crypto/sha256-js": ^5.0.0
+    "@aws-sdk/client-bedrock-agent-runtime": ^3.749.0
+    "@aws-sdk/client-bedrock-runtime": ^3.749.0
+    "@aws-sdk/client-dynamodb": ^3.749.0
+    "@aws-sdk/client-kendra": ^3.749.0
+    "@aws-sdk/client-lambda": ^3.749.0
+    "@aws-sdk/client-s3": ^3.749.0
+    "@aws-sdk/client-sagemaker-runtime": ^3.749.0
+    "@aws-sdk/client-sfn": ^3.749.0
+    "@aws-sdk/credential-provider-node": ^3.388.0
+    "@azure/search-documents": ^12.0.0
+    "@azure/storage-blob": ^12.15.0
+    "@browserbasehq/sdk": "*"
+    "@browserbasehq/stagehand": ^1.0.0
+    "@clickhouse/client": ^0.2.5
+    "@cloudflare/ai": "*"
+    "@datastax/astra-db-ts": ^1.0.0
+    "@elastic/elasticsearch": ^8.4.0
+    "@getmetal/metal-sdk": "*"
+    "@getzep/zep-cloud": ^1.0.6
+    "@getzep/zep-js": ^0.9.0
+    "@gomomento/sdk": ^1.51.1
+    "@gomomento/sdk-core": ^1.51.1
+    "@google-ai/generativelanguage": "*"
+    "@google-cloud/storage": ^6.10.1 || ^7.7.0
+    "@gradientai/nodejs-sdk": ^1.2.0
+    "@huggingface/inference": ^2.6.4
+    "@huggingface/transformers": ^3.2.3
+    "@ibm-cloud/watsonx-ai": "*"
+    "@lancedb/lancedb": ^0.12.0
+    "@langchain/core": ">=0.2.21 <0.4.0"
+    "@layerup/layerup-security": ^1.5.12
+    "@libsql/client": ^0.14.0
+    "@mendable/firecrawl-js": ^1.4.3
+    "@mlc-ai/web-llm": "*"
+    "@mozilla/readability": "*"
+    "@neondatabase/serverless": "*"
+    "@notionhq/client": ^2.2.10
+    "@opensearch-project/opensearch": "*"
+    "@pinecone-database/pinecone": "*"
+    "@planetscale/database": ^1.8.0
+    "@premai/prem-sdk": ^0.3.25
+    "@qdrant/js-client-rest": ^1.8.2
+    "@raycast/api": ^1.55.2
+    "@rockset/client": ^0.9.1
+    "@smithy/eventstream-codec": ^2.0.5
+    "@smithy/protocol-http": ^3.0.6
+    "@smithy/signature-v4": ^2.0.10
+    "@smithy/util-utf8": ^2.0.0
+    "@spider-cloud/spider-client": ^0.0.21
+    "@supabase/supabase-js": ^2.45.0
+    "@tensorflow-models/universal-sentence-encoder": "*"
+    "@tensorflow/tfjs-converter": "*"
+    "@tensorflow/tfjs-core": "*"
+    "@upstash/ratelimit": ^1.1.3 || ^2.0.3
+    "@upstash/redis": ^1.20.6
+    "@upstash/vector": ^1.1.1
+    "@vercel/kv": "*"
+    "@vercel/postgres": "*"
+    "@writerai/writer-sdk": ^0.40.2
+    "@xata.io/client": ^0.28.0
+    "@zilliz/milvus2-sdk-node": ">=2.3.5"
+    apify-client: ^2.7.1
+    assemblyai: ^4.6.0
+    better-sqlite3: ">=9.4.0 <12.0.0"
+    cassandra-driver: ^4.7.2
+    cborg: ^4.1.1
+    cheerio: ^1.0.0-rc.12
+    chromadb: "*"
+    closevector-common: 0.1.3
+    closevector-node: 0.1.6
+    closevector-web: 0.1.6
+    cohere-ai: "*"
+    convex: ^1.3.1
+    crypto-js: ^4.2.0
+    d3-dsv: ^2.0.0
+    discord.js: ^14.14.1
+    dria: ^0.0.3
+    duck-duck-scrape: ^2.2.5
+    epub2: ^3.0.1
+    fast-xml-parser: "*"
+    firebase-admin: ^11.9.0 || ^12.0.0
+    google-auth-library: "*"
+    googleapis: "*"
+    hnswlib-node: ^3.0.0
+    html-to-text: ^9.0.5
+    ibm-cloud-sdk-core: "*"
+    ignore: ^5.2.0
+    interface-datastore: ^8.2.11
+    ioredis: ^5.3.2
+    it-all: ^3.0.4
+    jsdom: "*"
+    jsonwebtoken: ^9.0.2
+    llmonitor: ^0.5.9
+    lodash: ^4.17.21
+    lunary: ^0.7.10
+    mammoth: ^1.6.0
+    mariadb: ^3.4.0
+    mongodb: ">=5.2.0"
+    mysql2: ^3.9.8
+    neo4j-driver: "*"
+    notion-to-md: ^3.1.0
+    officeparser: ^4.0.4
+    openai: "*"
+    pdf-parse: 1.1.1
+    pg: ^8.11.0
+    pg-copy-streams: ^6.0.5
+    pickleparser: ^0.2.1
+    playwright: ^1.32.1
+    portkey-ai: ^0.1.11
+    puppeteer: "*"
+    pyodide: ">=0.24.1 <0.27.0"
+    redis: "*"
+    replicate: "*"
+    sonix-speech-recognition: ^2.1.1
+    srt-parser-2: ^1.2.3
+    typeorm: ^0.3.20
+    typesense: ^1.5.3
+    usearch: ^1.1.1
+    voy-search: 0.6.2
+    weaviate-ts-client: "*"
+    web-auth-library: ^1.0.3
+    word-extractor: "*"
+    ws: ^8.14.2
+    youtubei.js: "*"
+  peerDependenciesMeta:
+    "@arcjet/redact":
+      optional: true
+    "@aws-crypto/sha256-js":
+      optional: true
+    "@aws-sdk/client-bedrock-agent-runtime":
+      optional: true
+    "@aws-sdk/client-bedrock-runtime":
+      optional: true
+    "@aws-sdk/client-dynamodb":
+      optional: true
+    "@aws-sdk/client-kendra":
+      optional: true
+    "@aws-sdk/client-lambda":
+      optional: true
+    "@aws-sdk/client-s3":
+      optional: true
+    "@aws-sdk/client-sagemaker-runtime":
+      optional: true
+    "@aws-sdk/client-sfn":
+      optional: true
+    "@aws-sdk/credential-provider-node":
+      optional: true
+    "@aws-sdk/dsql-signer":
+      optional: true
+    "@azure/search-documents":
+      optional: true
+    "@azure/storage-blob":
+      optional: true
+    "@browserbasehq/sdk":
+      optional: true
+    "@clickhouse/client":
+      optional: true
+    "@cloudflare/ai":
+      optional: true
+    "@datastax/astra-db-ts":
+      optional: true
+    "@elastic/elasticsearch":
+      optional: true
+    "@getmetal/metal-sdk":
+      optional: true
+    "@getzep/zep-cloud":
+      optional: true
+    "@getzep/zep-js":
+      optional: true
+    "@gomomento/sdk":
+      optional: true
+    "@gomomento/sdk-core":
+      optional: true
+    "@google-ai/generativelanguage":
+      optional: true
+    "@google-cloud/storage":
+      optional: true
+    "@gradientai/nodejs-sdk":
+      optional: true
+    "@huggingface/inference":
+      optional: true
+    "@huggingface/transformers":
+      optional: true
+    "@lancedb/lancedb":
+      optional: true
+    "@layerup/layerup-security":
+      optional: true
+    "@libsql/client":
+      optional: true
+    "@mendable/firecrawl-js":
+      optional: true
+    "@mlc-ai/web-llm":
+      optional: true
+    "@mozilla/readability":
+      optional: true
+    "@neondatabase/serverless":
+      optional: true
+    "@notionhq/client":
+      optional: true
+    "@opensearch-project/opensearch":
+      optional: true
+    "@pinecone-database/pinecone":
+      optional: true
+    "@planetscale/database":
+      optional: true
+    "@premai/prem-sdk":
+      optional: true
+    "@qdrant/js-client-rest":
+      optional: true
+    "@raycast/api":
+      optional: true
+    "@rockset/client":
+      optional: true
+    "@smithy/eventstream-codec":
+      optional: true
+    "@smithy/protocol-http":
+      optional: true
+    "@smithy/signature-v4":
+      optional: true
+    "@smithy/util-utf8":
+      optional: true
+    "@spider-cloud/spider-client":
+      optional: true
+    "@supabase/supabase-js":
+      optional: true
+    "@tensorflow-models/universal-sentence-encoder":
+      optional: true
+    "@tensorflow/tfjs-converter":
+      optional: true
+    "@tensorflow/tfjs-core":
+      optional: true
+    "@upstash/ratelimit":
+      optional: true
+    "@upstash/redis":
+      optional: true
+    "@upstash/vector":
+      optional: true
+    "@vercel/kv":
+      optional: true
+    "@vercel/postgres":
+      optional: true
+    "@writerai/writer-sdk":
+      optional: true
+    "@xata.io/client":
+      optional: true
+    "@zilliz/milvus2-sdk-node":
+      optional: true
+    apify-client:
+      optional: true
+    assemblyai:
+      optional: true
+    better-sqlite3:
+      optional: true
+    cassandra-driver:
+      optional: true
+    cborg:
+      optional: true
+    cheerio:
+      optional: true
+    chromadb:
+      optional: true
+    closevector-common:
+      optional: true
+    closevector-node:
+      optional: true
+    closevector-web:
+      optional: true
+    cohere-ai:
+      optional: true
+    convex:
+      optional: true
+    crypto-js:
+      optional: true
+    d3-dsv:
+      optional: true
+    discord.js:
+      optional: true
+    dria:
+      optional: true
+    duck-duck-scrape:
+      optional: true
+    epub2:
+      optional: true
+    fast-xml-parser:
+      optional: true
+    firebase-admin:
+      optional: true
+    google-auth-library:
+      optional: true
+    googleapis:
+      optional: true
+    hnswlib-node:
+      optional: true
+    html-to-text:
+      optional: true
+    ignore:
+      optional: true
+    interface-datastore:
+      optional: true
+    ioredis:
+      optional: true
+    it-all:
+      optional: true
+    jsdom:
+      optional: true
+    jsonwebtoken:
+      optional: true
+    llmonitor:
+      optional: true
+    lodash:
+      optional: true
+    lunary:
+      optional: true
+    mammoth:
+      optional: true
+    mariadb:
+      optional: true
+    mongodb:
+      optional: true
+    mysql2:
+      optional: true
+    neo4j-driver:
+      optional: true
+    notion-to-md:
+      optional: true
+    officeparser:
+      optional: true
+    pdf-parse:
+      optional: true
+    pg:
+      optional: true
+    pg-copy-streams:
+      optional: true
+    pickleparser:
+      optional: true
+    playwright:
+      optional: true
+    portkey-ai:
+      optional: true
+    puppeteer:
+      optional: true
+    pyodide:
+      optional: true
+    redis:
+      optional: true
+    replicate:
+      optional: true
+    sonix-speech-recognition:
+      optional: true
+    srt-parser-2:
+      optional: true
+    typeorm:
+      optional: true
+    typesense:
+      optional: true
+    usearch:
+      optional: true
+    voy-search:
+      optional: true
+    weaviate-ts-client:
+      optional: true
+    web-auth-library:
+      optional: true
+    word-extractor:
+      optional: true
+    ws:
+      optional: true
+    youtubei.js:
+      optional: true
+  checksum: 10c0/7945ba936c1b18506ca8ca70449d41208c627cadf52782a9d41e54e395fdc86a4421c01fa4977b582f7eddf5458dd7c418f932016c558a058abea3442828b8dd
+  languageName: node
+  linkType: hard
+
 "@langchain/core@npm:^0.3.26":
   version: 0.3.42
   resolution: "@langchain/core@npm:0.3.42"
@@ -1638,6 +2028,20 @@ __metadata:
   languageName: node
   linkType: hard
 
+"@langchain/openai@npm:>=0.2.0 <0.5.0":
+  version: 0.4.7
+  resolution: "@langchain/openai@npm:0.4.7"
+  dependencies:
+    js-tiktoken: "npm:^1.0.12"
+    openai: "npm:^4.87.3"
+    zod: "npm:^3.22.4"
+    zod-to-json-schema: "npm:^3.22.3"
+  peerDependencies:
+    "@langchain/core": ">=0.3.39 <0.4.0"
+  checksum: 10c0/4c809100c1e039c8c786141f097764cc399699ca9b149861396b2339147ad8b7fec5c97e5365a546bb7c123c4ec1943cfe3ef2eb5f056d6ab5a3fd1dc35940bc
+  languageName: node
+  linkType: hard
+
 "@langchain/openai@patch:@langchain/openai@npm%3A0.3.16#~/.yarn/patches/@langchain-openai-npm-0.3.16-e525b59526.patch":
   version: 0.3.16
   resolution: "@langchain/openai@patch:@langchain/openai@npm%3A0.3.16#~/.yarn/patches/@langchain-openai-npm-0.3.16-e525b59526.patch::version=0.3.16&hash=642f39"
@@ -2798,20 +3202,6 @@ __metadata:
   languageName: node
   linkType: hard
 
-"@types/command-line-args@npm:^5.2.3":
-  version: 5.2.3
-  resolution: "@types/command-line-args@npm:5.2.3"
-  checksum: 10c0/3a9bc58fd26e546391f6369dd28c03d59349dc4ac39eada1a5c39cc3578e02e4aac222615170e0db79b198ffba2af84fdbdda46e08c6edc4da42bc17ea85200f
-  languageName: node
-  linkType: hard
-
-"@types/command-line-usage@npm:^5.0.4":
-  version: 5.0.4
-  resolution: "@types/command-line-usage@npm:5.0.4"
-  checksum: 10c0/67840ebf4bcfee200c07d978669ad596fe2adc350fd5c19d44ec2248623575d96ec917f513d1d59453f8f57e879133861a4cc41c20045c07f6c959f1fcaac7ad
-  languageName: node
-  linkType: hard
-
 "@types/d3-color@npm:*":
   version: 3.1.3
   resolution: "@types/d3-color@npm:3.1.3"
@@ -3421,6 +3811,7 @@ __metadata:
     "@google/generative-ai": "npm:^0.21.0"
     "@hello-pangea/dnd": "npm:^16.6.0"
     "@kangfenmao/keyv-storage": "npm:^0.1.0"
+    "@langchain/community": "npm:^0.3.36"
     "@llm-tools/embedjs": "patch:@llm-tools/embedjs@npm%3A0.1.28#~/.yarn/patches/@llm-tools-embedjs-npm-0.1.28-8e4393fa2d.patch"
     "@llm-tools/embedjs-libsql": "npm:^0.1.28"
     "@llm-tools/embedjs-loader-csv": "npm:^0.1.28"
@@ -4091,6 +4482,13 @@ __metadata:
   languageName: node
   linkType: hard
 
+"binary-extensions@npm:^2.2.0":
+  version: 2.3.0
+  resolution: "binary-extensions@npm:2.3.0"
+  checksum: 10c0/75a59cafc10fb12a11d510e77110c6c7ae3f4ca22463d52487709ca7f18f69d886aa387557cc9864fbdb10153d0bdb4caacabf11541f55e89ed6e18d12ece2b5
+  languageName: node
+  linkType: hard
+
 "bindings@npm:^1.5.0":
   version: 1.5.0
   resolution: "bindings@npm:1.5.0"
@@ -6742,6 +7140,13 @@ __metadata:
   languageName: node
   linkType: hard
 
+"expr-eval@npm:^2.0.2":
+  version: 2.0.2
+  resolution: "expr-eval@npm:2.0.2"
+  checksum: 10c0/642f112ff28ea34574c595c3ad73ccd8e638498879a4dd28620c4dabebab2e11987a851266ba81883dae85a5800e0c93b3d06f81718b71a215f831534646e4f2
+  languageName: node
+  linkType: hard
+
 "express-rate-limit@npm:^7.5.0":
   version: 7.5.0
   resolution: "express-rate-limit@npm:7.5.0"
@@ -7095,6 +7500,15 @@ __metadata:
   languageName: node
   linkType: hard
 
+"flat@npm:^5.0.2":
+  version: 5.0.2
+  resolution: "flat@npm:5.0.2"
+  bin:
+    flat: cli.js
+  checksum: 10c0/f178b13482f0cd80c7fede05f4d10585b1f2fdebf26e12edc138e32d3150c6ea6482b7f12813a1091143bad52bb6d3596bca51a162257a21163c0ff438baa5fe
+  languageName: node
+  linkType: hard
+
 "flatted@npm:^3.2.9":
   version: 3.3.3
   resolution: "flatted@npm:3.3.3"
@@ -9067,7 +9481,7 @@ __metadata:
   languageName: node
   linkType: hard
 
-"langchain@npm:^0.3.8":
+"langchain@npm:>=0.2.3 <0.3.0 || >=0.3.4 <0.4.0, langchain@npm:^0.3.8":
   version: 0.3.19
   resolution: "langchain@npm:0.3.19"
   dependencies:
@@ -16029,6 +16443,15 @@ __metadata:
   languageName: node
   linkType: hard
 
+"zod-to-json-schema@npm:^3.22.5":
+  version: 3.24.5
+  resolution: "zod-to-json-schema@npm:3.24.5"
+  peerDependencies:
+    zod: ^3.24.1
+  checksum: 10c0/0745b94ba53e652d39f262641cdeb2f75d24339fb6076a38ce55bcf53d82dfaea63adf524ebc5f658681005401687f8e9551c4feca7c4c882e123e66091dfb90
+  languageName: node
+  linkType: hard
+
 "zod-validation-error@npm:^3.4.0":
   version: 3.4.0
   resolution: "zod-validation-error@npm:3.4.0"
@@ -16038,7 +16461,7 @@ __metadata:
   languageName: node
   linkType: hard
 
-"zod@npm:^3.22.4, zod@npm:^3.23.8":
+"zod@npm:^3.22.3, zod@npm:^3.22.4, zod@npm:^3.23.8":
   version: 3.24.2
   resolution: "zod@npm:3.24.2"
   checksum: 10c0/c638c7220150847f13ad90635b3e7d0321b36cce36f3fc6050ed960689594c949c326dfe2c6fa87c14b126ee5d370ccdebd6efb304f41ef5557a4aaca2824565