@smythos/sre 1.5.43 → 1.5.45

package/src/types/LLM.types.ts

@@ -1,491 +1,496 @@
-import OpenAI from 'openai';
-import Anthropic from '@anthropic-ai/sdk';
-import { FunctionCallingMode, ModelParams, GenerateContentRequest } from '@google/generative-ai';
-
-import { BinaryInput } from '@sre/helpers/BinaryInput.helper';
-import { type models } from '@sre/LLMManager/models';
-import { AccessRequest } from '@sre/Security/AccessControl/AccessRequest.class';
-import { ConverseCommandInput } from '@aws-sdk/client-bedrock-runtime';
-
-export type LLMProvider = Extract<(typeof models)[keyof typeof models], { llm: string }>['llm'] | 'VertexAI' | 'Bedrock';
-export type LLMModel = keyof typeof models;
-export type LLMModelInfo = (typeof models)[LLMModel];
-
-// Google Service Account Credentials Interface
-export interface VertexAICredentials {
-    type: 'service_account';
-    project_id: string;
-    private_key_id: string;
-    private_key: string;
-    client_email: string;
-    client_id: string;
-    auth_uri: string;
-    token_uri: string;
-    auth_provider_x509_cert_url: string;
-    client_x509_cert_url: string;
-    universe_domain?: string; // Optional, defaults to "googleapis.com"
-}
-
-// Basic LLM Credentials Interface
-export interface BasicCredentials {
-    apiKey: string;
-    isUserKey: boolean;
-}
-
-// AWS Bedrock Credentials Interface
-export interface BedrockCredentials {
-    accessKeyId: string;
-    secretAccessKey: string;
-    sessionToken?: string;
-}
-
-// Union type for all credential types
-export type ILLMConnectorCredentials = BasicCredentials | BedrockCredentials | VertexAICredentials;
-
-export type TOpenAIResponseToolChoice = OpenAI.Responses.ToolChoiceOptions | OpenAI.Responses.ToolChoiceTypes | OpenAI.Responses.ToolChoiceFunction;
-export type TLLMToolChoice = OpenAI.ChatCompletionToolChoiceOption;
-
-export type TOpenAIToolsInfo = {
-    webSearch: {
-        enabled: boolean;
-        contextSize: TSearchContextSize;
-        city?: string;
-        country?: string;
-        region?: string;
-        timezone?: string;
-    };
-};
-
-export type TxAIToolsInfo = {
-    search: {
-        enabled: boolean;
-        mode?: 'auto' | 'on' | 'off';
-        returnCitations?: boolean;
-        maxResults?: number;
-        dataSources?: string[];
-        country?: string;
-        excludedWebsites?: string[];
-        allowedWebsites?: string[];
-        includedXHandles?: string[];
-        excludedXHandles?: string[];
-        postFavoriteCount?: number;
-        postViewCount?: number;
-        rssLinks?: string;
-        safeSearch?: boolean;
-        fromDate?: string;
-        toDate?: string;
-    };
-};
-
-export type TToolsInfo = {
-    openai: TOpenAIToolsInfo;
-    xai: TxAIToolsInfo;
-};
-
-export type TSearchContextSize = 'low' | 'medium' | 'high';
-
-export type TLLMParams = {
-    model: TLLMModel | string;
-
-    prompt?: string;
-    messages?: any[]; // TODO [Forhad]: apply proper typing
-    temperature?: number;
-    maxTokens?: number;
-    stopSequences?: string[];
-    topP?: number;
-    topK?: number;
-    frequencyPenalty?: number;
-    presencePenalty?: number;
-    responseFormat?: any; // TODO [Forhad]: apply proper typing
-    modelInfo?: TCustomLLMModel;
-    files?: BinaryInput[];
-    toolsConfig?: {
-        tools?: OpenAI.ChatCompletionTool[] | OpenAI.Responses.Tool[] | OpenAI.Responses.WebSearchTool[];
-        tool_choice?: TLLMToolChoice;
-    };
-    baseURL?: string;
-
-    size?: OpenAI.Images.ImageGenerateParams['size'] | OpenAI.Images.ImageEditParams['size']; // for image generation and image editing
-    quality?: 'standard' | 'hd'; // for image generation
-    n?: number; // for image generation
-    style?: 'vivid' | 'natural'; // for image generation
-
-    cache?: boolean;
-    agentId?: string;
-    teamId?: string;
-    thinking?: {
-        // for Anthropic
-        type: 'enabled' | 'disabled';
-        budget_tokens: number;
-    };
-    maxThinkingTokens?: number;
-
-    // #region Search
-    // Web search parameters (will be organized into toolsInfo.webSearch internally)
-    useWebSearch?: boolean;
-    webSearchContextSize?: TSearchContextSize;
-    webSearchCity?: string;
-    webSearchCountry?: string;
-    webSearchRegion?: string;
-    webSearchTimezone?: string;
-
-    // xAI specific search parameters (consider moving to toolsInfo.xaiSearch)
-    useSearch?: boolean;
-    searchMode?: 'auto' | 'on' | 'off';
-    returnCitations?: boolean;
-    maxSearchResults?: number;
-    searchDataSources?: string[];
-    searchCountry?: string;
-    excludedWebsites?: string[];
-    allowedWebsites?: string[];
-    includedXHandles?: string[];
-    excludedXHandles?: string[];
-    postFavoriteCount?: number;
-    postViewCount?: number;
-    rssLinks?: string;
-    safeSearch?: boolean;
-    fromDate?: string;
-    toDate?: string;
-    // #endregion
-
-    useReasoning?: boolean;
-    /**
-     * Controls the level of effort the model will put into reasoning
-     * For GPT-OSS models (20B, 120B): "low" | "medium" | "high"
-     * For Qwen 3 32B: "none" | "default"
-     */
-    reasoningEffort?: 'none' | 'default' | 'low' | 'medium' | 'high';
-    max_output_tokens?: number;
-    abortSignal?: AbortSignal;
-};
-
-export type TLLMPreparedParams = TLLMParams & {
-    body: any;
-    modelEntryName?: string; // for usage reporting
-    credentials?: ILLMConnectorCredentials;
-    isUserKey?: boolean;
-    capabilities?: {
-        search?: boolean;
-        reasoning?: boolean;
-        imageGeneration?: boolean;
-        imageEditing?: boolean;
-    };
-    toolsInfo?: TToolsInfo;
-};
-
-export type TLLMConnectorParams = Omit<TLLMParams, 'model'> & {
-    //the LLMConnector accepts a model object that we extract the model info from instead of relying on the internal models list
-    model: string | TLLMModel | TCustomLLMModel;
-};
-
-export type TLLMModelEntry = {
-    llm: string;
-    tokens?: number;
-    completionTokens?: number;
-    enabled?: boolean;
-    components?: string[];
-    alias?: string;
-    tags?: string[];
-    keyOptions?: {
-        tokens: number;
-        completionTokens: number;
-    };
-};
-
-export enum TLLMCredentials {
-    Vault = 'vault',
-    Internal = 'internal',
-    BedrockVault = 'bedrock_vault',
-    VertexAIVault = 'vertexai_vault',
-    None = 'none',
-}
-export type TLLMModel = {
-    llm?: string;
-    isCustomLLM?: boolean;
-    modelId?: string;
-    tokens?: number;
-    completionTokens?: number;
-    components?: string[];
-    tags?: string[];
-    label?: string;
-    provider?: LLMProvider;
-    features?: string[];
-    enabled?: boolean;
-    alias?: string;
-    baseURL?: string;
-    keyOptions?: {
-        tokens: number;
-        completionTokens: number;
-    };
-    credentials?: TLLMCredentials;
-
-    //models can come with predefined params
-    //this can also be used to pass a preconfigured model object
-    params?: TLLMParams;
-    /**
-     * Specifies the API interface type to use for this model
-     * Examples: 'chat.completions', 'responses'
-     * This determines which OpenAI API endpoint and interface implementation to use
-     */
-    interface?: 'chat.completions' | 'responses';
-};
-
-// #region [ Handle extendable LLM Providers ] ================================================
-export const BuiltinLLMProviders = {
-    Echo: 'Echo',
-    OpenAI: 'OpenAI',
-    DeepSeek: 'DeepSeek',
-    GoogleAI: 'GoogleAI',
-    Anthropic: 'Anthropic',
-    Groq: 'Groq',
-    TogetherAI: 'TogetherAI',
-    Bedrock: 'Bedrock',
-    VertexAI: 'VertexAI',
-    xAI: 'xAI',
-    Perplexity: 'Perplexity',
-} as const;
-// Base provider type
-export type TBuiltinLLMProvider = (typeof BuiltinLLMProviders)[keyof typeof BuiltinLLMProviders];
-
-// Extensible interface for custom providers
-export interface ILLMProviders {}
-// Combined provider type that can be extended
-export type TLLMProvider = TBuiltinLLMProvider | keyof ILLMProviders;
-
-// For backward compatibility, export the built-in providers as enum-like object
-export const TLLMProvider = BuiltinLLMProviders;
-
-// #endregion
-
-export type TBedrockSettings = {
-    keyIDName: string;
-    secretKeyName: string;
-    sessionKeyName: string;
-};
-export type TVertexAISettings = {
-    projectId: string;
-    credentialsName: string;
-    jsonCredentialsName: string;
-    apiEndpoint?: string;
-};
-
-export type TCustomLLMModel = TLLMModel & {
-    name: string;
-    settings: {
-        foundationModel: string;
-        customModel: string;
-        region: string;
-    } & (TBedrockSettings | TVertexAISettings);
-};
-
-//#region === LLM Tools ===========================
-export type ToolData = {
-    index: number;
-    id: string;
-    type: string;
-    name: string;
-    arguments: string | Record<string, any>;
-    role: 'user' | 'tool' | 'assistant';
-    result?: string; // result string from the used tool
-    function?: any;
-    error?: string; // for Bedrock
-};
-
-/**
- * Base tool definition interface - only truly common properties
- * All provider-specific tool definitions extend from this
- */
-export interface ToolDefinition {
-    name: string;
-    description: string;
-}
-
-/**
- * OpenAI-specific tool definition
- * Extends base with OpenAI's parameter format
- */
-export interface OpenAIToolDefinition extends ToolDefinition {
-    parameters: {
-        type: 'object';
-        properties: Record<string, unknown>;
-        required?: string[];
-    };
-}
-
-/**
- * Anthropic-specific tool definition
- * Extends base with Anthropic's input_schema format
- */
-export interface AnthropicToolDefinition extends ToolDefinition {
-    input_schema: {
-        type: 'object';
-        properties: Record<string, unknown>;
-        required: string[];
-    };
-}
-
-/**
- * Legacy tool definition for backward compatibility
- * @deprecated Use provider-specific definitions instead
- */
-export interface LegacyToolDefinition extends ToolDefinition {
-    properties?: Record<string, unknown>;
-    requiredFields?: string[];
-}
-export type ToolChoice = OpenAI.ChatCompletionToolChoiceOption | FunctionCallingMode;
-
-export interface ToolsConfig {
-    tools?: ToolDefinition[];
-    tool_choice?: ToolChoice;
-}
-
-//#endregion
-
-export enum TLLMMessageRole {
-    User = 'user',
-    Assistant = 'assistant',
-    System = 'system',
-    Model = 'model',
-    Tool = 'tool',
-    Function = 'function',
-}
-
-export type TLLMMessageBlock = {
-    role: TLLMMessageRole;
-    content?:
-        | string
-        | { text: string }[]
-        | Array<Anthropic.TextBlockParam | Anthropic.ImageBlockParam | Anthropic.ToolUseBlockParam | Anthropic.ToolResultBlockParam>;
-    parts?: {
-        text?: string;
-        functionCall?: { name: string; args: string };
-        functionResponse?: { name: string; response: { name: string; content: string } };
-    }[]; // for Google Vertex AI
-    tool_calls?: ToolData[];
-};
-
-export type TLLMToolResultMessageBlock = TLLMMessageBlock & {
-    tool_call_id?: string; // for tool result message block of OpenAI
-    name?: string; // for tool result message block of OpenAI
-};
-
-export type GenerateImageConfig = {
-    size?: '256x256' | '512x512' | '1024x1024' | '1792x1024' | '1024x1792';
-    quality?: 'standard' | 'hd';
-    model: string;
-    style?: 'vivid' | 'natural';
-    n?: number;
-    response_format?: 'url' | 'b64_json';
-    prompt?: string;
-};
-
-// ! Deprecated
-export type TLLMInputMessage = {
-    role: string;
-    content?: string | { text: string }[];
-    parts?: { text: string }[]; // * 'part' is for Google Vertex AI
-};
-
-export interface ILLMContextStore {
-    save(messages: any[]): Promise<void>;
-    load(count?: number): Promise<any[]>;
-    getMessage(message_id: string): Promise<any[]>;
-}
-
-export enum APIKeySource {
-    Smyth = 'smyth-managed',
-    User = 'user-managed',
-    Custom = 'custom',
-}
-
-export interface SmythLLMUsage {
-    sourceId: string;
-    input_tokens: number;
-    input_tokens_cache_write: number;
-    input_tokens_cache_read: number;
-    output_tokens: number;
-    keySource?: APIKeySource;
-    agentId: string;
-    teamId: string;
-    tier?: string; // for Google AI
-}
-
-export interface SmythTaskUsage {
-    sourceId: string;
-    number: number;
-    agentId: string;
-    teamId: string;
-}
-
-export type TLLMModelsList = {
-    [key: string]: TLLMModel | TCustomLLMModel;
-};
-
-export enum TLLMEvent {
-    /** Generated response chunks */
-    Content = 'content',
-    /** Thinking blocks/chunks */
-    Thinking = 'thinking',
-    /** End of the response */
-    End = 'end',
-    /** Error */
-    Error = 'error',
-    /** Tool information : emitted by the LLM determines the next tool call */
-    ToolInfo = 'toolInfo',
-    /** Tool call : emitted before the tool call */
-    ToolCall = 'toolCall',
-    /** Tool result : emitted after the tool call */
-    ToolResult = 'toolResult',
-    /** Tokens usage information */
-    Usage = 'usage',
-    /** Interrupted : emitted when the response is interrupted before completion */
-    Interrupted = 'interrupted',
-}
-
-export interface ILLMRequestContext {
-    modelEntryName: string;
-    agentId: string;
-    teamId: string;
-    isUserKey: boolean;
-    hasFiles?: boolean;
-    modelInfo: TCustomLLMModel | TLLMModel;
-    credentials: ILLMConnectorCredentials;
-    toolsInfo?: TToolsInfo;
-}
-
-// Generic interface that can be extended by specific providers
-export interface ILLMRequestFuncParams<TBody = any> {
-    acRequest: AccessRequest;
-    body: TBody;
-    context: ILLMRequestContext;
-}
-
-// For future providers, you can add similar types:
-// export type TAnthropicRequestBody = Anthropic.MessageCreateParams | Anthropic.MessageStreamParams;
-// export type IAnthropicRequestFuncParams = ILLMRequestFuncParams<TAnthropicRequestBody>;
-
-export type TLLMChatResponse = {
-    content: string;
-    finishReason: string;
-    thinkingContent?: string;
-    usage?: any;
-    useTool?: boolean;
-    toolsData?: ToolData[];
-    message?: OpenAI.ChatCompletionMessageParam | Anthropic.MessageParam;
-};
-
-export type TOpenAIRequestBody =
-    | OpenAI.ChatCompletionCreateParamsNonStreaming
-    | OpenAI.ChatCompletionCreateParamsStreaming
-    | OpenAI.ChatCompletionCreateParams
-    | OpenAI.Responses.ResponseCreateParams
-    | OpenAI.Responses.ResponseCreateParamsNonStreaming
-    | OpenAI.Responses.ResponseCreateParamsStreaming
-    | OpenAI.Images.ImageGenerateParams
-    | OpenAI.Images.ImageEditParams;
-
-export type TAnthropicRequestBody = Anthropic.MessageCreateParamsNonStreaming;
-
-export type TGoogleAIRequestBody = ModelParams & { messages: string | TLLMMessageBlock[] | GenerateContentRequest };
-
-export type TLLMRequestBody = TOpenAIRequestBody | TAnthropicRequestBody | TGoogleAIRequestBody | ConverseCommandInput;
+import OpenAI from 'openai';
+import Anthropic from '@anthropic-ai/sdk';
+import { FunctionCallingMode, ModelParams, GenerateContentRequest } from '@google/generative-ai';
+
+import { BinaryInput } from '@sre/helpers/BinaryInput.helper';
+import { type models } from '@sre/LLMManager/models';
+import { AccessRequest } from '@sre/Security/AccessControl/AccessRequest.class';
+import { ConverseCommandInput } from '@aws-sdk/client-bedrock-runtime';
+
+export type LLMProvider = Extract<(typeof models)[keyof typeof models], { llm: string }>['llm'] | 'VertexAI' | 'Bedrock';
+export type LLMModel = keyof typeof models;
+export type LLMModelInfo = (typeof models)[LLMModel];
+
+// Google Service Account Credentials Interface
+export interface VertexAICredentials {
+    type: 'service_account';
+    project_id: string;
+    private_key_id: string;
+    private_key: string;
+    client_email: string;
+    client_id: string;
+    auth_uri: string;
+    token_uri: string;
+    auth_provider_x509_cert_url: string;
+    client_x509_cert_url: string;
+    universe_domain?: string; // Optional, defaults to "googleapis.com"
+}
+
+// Basic LLM Credentials Interface
+export interface BasicCredentials {
+    apiKey: string;
+    isUserKey: boolean;
+}
+
+// AWS Bedrock Credentials Interface
+export interface BedrockCredentials {
+    accessKeyId: string;
+    secretAccessKey: string;
+    sessionToken?: string;
+}
+
+// Union type for all credential types
+export type ILLMConnectorCredentials = BasicCredentials | BedrockCredentials | VertexAICredentials;
+
+export type TOpenAIResponseToolChoice = OpenAI.Responses.ToolChoiceOptions | OpenAI.Responses.ToolChoiceTypes | OpenAI.Responses.ToolChoiceFunction;
+export type TLLMToolChoice = OpenAI.ChatCompletionToolChoiceOption;
+
+// Local alias to the upstream OpenAI reasoning effort union type
+export type OpenAIReasoningEffort = NonNullable<OpenAI.Responses.ResponseCreateParams['reasoning']>['effort'];
+
+export type TOpenAIToolsInfo = {
+    webSearch: {
+        enabled: boolean;
+        contextSize: TSearchContextSize;
+        city?: string;
+        country?: string;
+        region?: string;
+        timezone?: string;
+    };
+};
+
+export type TxAIToolsInfo = {
+    search: {
+        enabled: boolean;
+        mode?: 'auto' | 'on' | 'off';
+        returnCitations?: boolean;
+        maxResults?: number;
+        dataSources?: string[];
+        country?: string;
+        excludedWebsites?: string[];
+        allowedWebsites?: string[];
+        includedXHandles?: string[];
+        excludedXHandles?: string[];
+        postFavoriteCount?: number;
+        postViewCount?: number;
+        rssLinks?: string;
+        safeSearch?: boolean;
+        fromDate?: string;
+        toDate?: string;
+    };
+};
+
+export type TToolsInfo = {
+    openai: TOpenAIToolsInfo;
+    xai: TxAIToolsInfo;
+};
+
+export type TSearchContextSize = 'low' | 'medium' | 'high';
+
+export type TLLMParams = {
+    model: TLLMModel | string;
+
+    prompt?: string;
+    messages?: any[]; // TODO [Forhad]: apply proper typing
+    temperature?: number;
+    maxTokens?: number;
+    stopSequences?: string[];
+    topP?: number;
+    topK?: number;
+    frequencyPenalty?: number;
+    presencePenalty?: number;
+    responseFormat?: any; // TODO [Forhad]: apply proper typing
+    modelInfo?: TCustomLLMModel;
+    files?: BinaryInput[];
+    toolsConfig?: {
+        tools?: OpenAI.ChatCompletionTool[] | OpenAI.Responses.Tool[] | OpenAI.Responses.WebSearchTool[];
+        tool_choice?: TLLMToolChoice;
+    };
+    baseURL?: string;
+
+    size?: OpenAI.Images.ImageGenerateParams['size'] | OpenAI.Images.ImageEditParams['size']; // for image generation and image editing
+    quality?: 'standard' | 'hd'; // for image generation
+    n?: number; // for image generation
+    style?: 'vivid' | 'natural'; // for image generation
+
+    cache?: boolean;
+    agentId?: string;
+    teamId?: string;
+    thinking?: {
+        // for Anthropic
+        type: 'enabled' | 'disabled';
+        budget_tokens: number;
+    };
+    maxThinkingTokens?: number;
+
+    // #region Search
+    // Web search parameters (will be organized into toolsInfo.webSearch internally)
+    useWebSearch?: boolean;
+    webSearchContextSize?: TSearchContextSize;
+    webSearchCity?: string;
+    webSearchCountry?: string;
+    webSearchRegion?: string;
+    webSearchTimezone?: string;
+
+    // xAI specific search parameters (consider moving to toolsInfo.xaiSearch)
+    useSearch?: boolean;
+    searchMode?: 'auto' | 'on' | 'off';
+    returnCitations?: boolean;
+    maxSearchResults?: number;
+    searchDataSources?: string[];
+    searchCountry?: string;
+    excludedWebsites?: string[];
+    allowedWebsites?: string[];
+    includedXHandles?: string[];
+    excludedXHandles?: string[];
+    postFavoriteCount?: number;
+    postViewCount?: number;
+    rssLinks?: string;
+    safeSearch?: boolean;
+    fromDate?: string;
+    toDate?: string;
+    // #endregion
+
+    useReasoning?: boolean;
+    /**
+     * Controls the level of effort the model will put into reasoning
+     * For GPT-OSS models (20B, 120B): "low" | "medium" | "high"
+     * For Qwen 3 32B: "none" | "default"
+     */
+    reasoningEffort?: 'none' | 'default' | OpenAIReasoningEffort;
+    max_output_tokens?: number;
+    verbosity?: OpenAI.Responses.ResponseCreateParams['text']['verbosity'];
+    abortSignal?: AbortSignal;
+};
+
+export type TLLMPreparedParams = TLLMParams & {
+    body: any;
+    modelEntryName?: string; // for usage reporting
+    credentials?: ILLMConnectorCredentials;
+    isUserKey?: boolean;
+    capabilities?: {
+        search?: boolean;
+        reasoning?: boolean;
+        imageGeneration?: boolean;
+        imageEditing?: boolean;
+    };
+    toolsInfo?: TToolsInfo;
+};
+
+export type TLLMConnectorParams = Omit<TLLMParams, 'model'> & {
+    //the LLMConnector accepts a model object that we extract the model info from instead of relying on the internal models list
+    model: string | TLLMModel | TCustomLLMModel;
+};
+
+export type TLLMModelEntry = {
+    llm: string;
+    tokens?: number;
+    completionTokens?: number;
+    enabled?: boolean;
+    components?: string[];
+    alias?: string;
+    tags?: string[];
+    keyOptions?: {
+        tokens: number;
+        completionTokens: number;
+    };
+};
+
+export enum TLLMCredentials {
+    Vault = 'vault',
+    Internal = 'internal',
+    BedrockVault = 'bedrock_vault',
+    VertexAIVault = 'vertexai_vault',
+    None = 'none',
+}
+export type TLLMModel = {
+    llm?: string;
+    isCustomLLM?: boolean;
+    modelId?: string;
+    tokens?: number;
+    completionTokens?: number;
+    components?: string[];
+    tags?: string[];
+    label?: string;
+    provider?: LLMProvider;
+    features?: string[];
+    enabled?: boolean;
+    alias?: string;
+    baseURL?: string;
+    keyOptions?: {
+        tokens: number;
+        completionTokens: number;
+    };
+    credentials?: TLLMCredentials;
+
+    //models can come with predefined params
+    //this can also be used to pass a preconfigured model object
+    params?: TLLMParams;
+    /**
+     * Specifies the API interface type to use for this model
+     * Examples: 'chat.completions', 'responses'
+     * This determines which OpenAI API endpoint and interface implementation to use
+     */
+    interface?: 'chat.completions' | 'responses';
+};
+
+// #region [ Handle extendable LLM Providers ] ================================================
+export const BuiltinLLMProviders = {
+    Echo: 'Echo',
+    OpenAI: 'OpenAI',
+    DeepSeek: 'DeepSeek',
+    GoogleAI: 'GoogleAI',
+    Anthropic: 'Anthropic',
+    Groq: 'Groq',
+    TogetherAI: 'TogetherAI',
+    Bedrock: 'Bedrock',
+    VertexAI: 'VertexAI',
+    xAI: 'xAI',
+    Perplexity: 'Perplexity',
+} as const;
+// Base provider type
+export type TBuiltinLLMProvider = (typeof BuiltinLLMProviders)[keyof typeof BuiltinLLMProviders];
+
+// Extensible interface for custom providers
+export interface ILLMProviders {}
+// Combined provider type that can be extended
+export type TLLMProvider = TBuiltinLLMProvider | keyof ILLMProviders;
+
+// For backward compatibility, export the built-in providers as enum-like object
+export const TLLMProvider = BuiltinLLMProviders;
+
+// #endregion
+
+export type TBedrockSettings = {
+    keyIDName: string;
+    secretKeyName: string;
+    sessionKeyName: string;
+};
+export type TVertexAISettings = {
+    projectId: string;
+    credentialsName: string;
+    jsonCredentialsName: string;
+    apiEndpoint?: string;
+};
+
+export type TCustomLLMModel = TLLMModel & {
+    name: string;
+    settings: {
+        foundationModel: string;
+        customModel: string;
+        region: string;
+    } & (TBedrockSettings | TVertexAISettings);
+};
+
+//#region === LLM Tools ===========================
+export type ToolData = {
+    index: number;
+    id: string;
+    type: string;
+    name: string;
+    arguments: string | Record<string, any>;
+    role: 'user' | 'tool' | 'assistant';
+    result?: string; // result string from the used tool
+    function?: any;
+    error?: string; // for Bedrock
+    callId?: string; // for OpenAI Responses API call ID mapping
+};
+
+/**
+ * Base tool definition interface - only truly common properties
+ * All provider-specific tool definitions extend from this
+ */
+export interface ToolDefinition {
+    name: string;
+    description: string;
+}
+
+/**
+ * OpenAI-specific tool definition
+ * Extends base with OpenAI's parameter format
+ */
+export interface OpenAIToolDefinition extends ToolDefinition {
+    parameters: {
+        type: 'object';
+        properties: Record<string, unknown>;
+        required?: string[];
+    };
+}
+
+/**
+ * Anthropic-specific tool definition
+ * Extends base with Anthropic's input_schema format
+ */
+export interface AnthropicToolDefinition extends ToolDefinition {
+    input_schema: {
+        type: 'object';
+        properties: Record<string, unknown>;
+        required: string[];
+    };
+}
+
+/**
+ * Legacy tool definition for backward compatibility
+ * @deprecated Use provider-specific definitions instead
+ */
+export interface LegacyToolDefinition extends ToolDefinition {
+    properties?: Record<string, unknown>;
+    requiredFields?: string[];
+}
+export type ToolChoice = OpenAI.ChatCompletionToolChoiceOption | FunctionCallingMode;
+
+export interface ToolsConfig {
+    tools?: ToolDefinition[];
+    tool_choice?: ToolChoice;
+}
+
+//#endregion
+
+export enum TLLMMessageRole {
+    User = 'user',
+    Assistant = 'assistant',
+    System = 'system',
+    Model = 'model',
+    Tool = 'tool',
+    Function = 'function',
+}
+
+export type TLLMMessageBlock = {
+    role: TLLMMessageRole;
+    content?:
+        | string
+        | { text: string }[]
+        | Array<Anthropic.TextBlockParam | Anthropic.ImageBlockParam | Anthropic.ToolUseBlockParam | Anthropic.ToolResultBlockParam>;
+    parts?: {
+        text?: string;
+        functionCall?: { name: string; args: string };
+        functionResponse?: { name: string; response: { name: string; content: string } };
+    }[]; // for Google Vertex AI
+    tool_calls?: ToolData[];
+};
+
+export type TLLMToolResultMessageBlock = TLLMMessageBlock & {
+    tool_call_id?: string; // for tool result message block of OpenAI
+    name?: string; // for tool result message block of OpenAI
+};
+
+export type GenerateImageConfig = {
+    size?: '256x256' | '512x512' | '1024x1024' | '1792x1024' | '1024x1792';
+    quality?: 'standard' | 'hd';
+    model: string;
+    style?: 'vivid' | 'natural';
+    n?: number;
+    response_format?: 'url' | 'b64_json';
+    prompt?: string;
+};
+
+// ! Deprecated
+export type TLLMInputMessage = {
+    role: string;
+    content?: string | { text: string }[];
+    parts?: { text: string }[]; // * 'part' is for Google Vertex AI
+};
+
+export interface ILLMContextStore {
+    save(messages: any[]): Promise<void>;
+    load(count?: number): Promise<any[]>;
+    getMessage(message_id: string): Promise<any[]>;
+}
+
+export enum APIKeySource {
+    Smyth = 'smyth-managed',
+    User = 'user-managed',
+    Custom = 'custom',
+}
+
+export interface SmythLLMUsage {
+    sourceId: string;
+    input_tokens: number;
+    input_tokens_cache_write: number;
+    input_tokens_cache_read: number;
+    output_tokens: number;
+    keySource?: APIKeySource;
+    agentId: string;
+    teamId: string;
+    tier?: string; // for Google AI
+}
+
+export interface SmythTaskUsage {
+    sourceId: string;
+    number: number;
+    agentId: string;
+    teamId: string;
+}
+
+export type TLLMModelsList = {
+    [key: string]: TLLMModel | TCustomLLMModel;
+};
+
+export enum TLLMEvent {
+    /** Generated response chunks */
+    Content = 'content',
+    /** Thinking blocks/chunks */
+    Thinking = 'thinking',
+    /** End of the response */
+    End = 'end',
+    /** Error */
+    Error = 'error',
+    /** Tool information : emitted by the LLM determines the next tool call */
+    ToolInfo = 'toolInfo',
+    /** Tool call : emitted before the tool call */
+    ToolCall = 'toolCall',
+    /** Tool result : emitted after the tool call */
+    ToolResult = 'toolResult',
+    /** Tokens usage information */
+    Usage = 'usage',
+    /** Interrupted : emitted when the response is interrupted before completion */
+    Interrupted = 'interrupted',
+}
+
+export interface ILLMRequestContext {
+    modelEntryName: string;
+    agentId: string;
+    teamId: string;
+    isUserKey: boolean;
+    hasFiles?: boolean;
+    modelInfo: TCustomLLMModel | TLLMModel;
+    credentials: ILLMConnectorCredentials;
+    toolsInfo?: TToolsInfo;
+}
+
+// Generic interface that can be extended by specific providers
+export interface ILLMRequestFuncParams<TBody = any> {
+    acRequest: AccessRequest;
+    body: TBody;
+    context: ILLMRequestContext;
+}
+
+// For future providers, you can add similar types:
+// export type TAnthropicRequestBody = Anthropic.MessageCreateParams | Anthropic.MessageStreamParams;
+// export type IAnthropicRequestFuncParams = ILLMRequestFuncParams<TAnthropicRequestBody>;
+
+export type TLLMChatResponse = {
+    content: string;
+    finishReason: string;
+    thinkingContent?: string;
+    usage?: any;
+    useTool?: boolean;
+    toolsData?: ToolData[];
+    message?: OpenAI.ChatCompletionMessageParam | Anthropic.MessageParam;
+};
+
+export type TOpenAIRequestBody =
+    | OpenAI.ChatCompletionCreateParamsNonStreaming
+    | OpenAI.ChatCompletionCreateParamsStreaming
+    | OpenAI.ChatCompletionCreateParams
+    | OpenAI.Responses.ResponseCreateParams
+    | OpenAI.Responses.ResponseCreateParamsNonStreaming
+    | OpenAI.Responses.ResponseCreateParamsStreaming
+    | OpenAI.Images.ImageGenerateParams
+    | OpenAI.Images.ImageEditParams;
+
+export type TAnthropicRequestBody = Anthropic.MessageCreateParamsNonStreaming;
+
+export type TGoogleAIRequestBody = ModelParams & { messages: string | TLLMMessageBlock[] | GenerateContentRequest };
+
+export type TLLMRequestBody = TOpenAIRequestBody | TAnthropicRequestBody | TGoogleAIRequestBody | ConverseCommandInput;