@agentforce/adk/lib/provider/openrouter.ts

AgentForce Agent Development Kit - A powerful framework for building AI agents and servers

import OpenAI from "openai";
import type { ChatCompletionMessageParam } from "openai/resources/chat/completions";
import type { Tool } from "../types";
import { OpenRouterToolUse } from "./openrouter-tooluse";
import type { AgentForceLogger, ModelConfig } from "../types";

/**
 * Interface for OpenRouter provider functionality
 * @interface OpenRouterProviderInterface
 * @property {function} generate - Generate response from prompt without tools
 * @property {function} generateWithTools - Generate response with tool support
 * @property {function} chatWithTools - Chat with tool support using message history
 * @property {function} chat - Chat without tool support using message history
 * @property {function} getModel - Get current model name
 * @property {function} setModel - Set model name
 */
export interface OpenRouterProviderInterface {
    generate(prompt: string, system?: string): Promise<string>;
    generateWithTools(prompt: string, tools: Tool[], system?: string, logger?: AgentForceLogger, agent?: any): Promise<string>;
    chatWithTools(messages: Array<{ role: string; content: string }>, tools: Tool[], logger?: AgentForceLogger, agent?: any): Promise<string>;
    chat(messages: Array<{ role: string; content: string }>): Promise<string>;
    getModel(): string;
    setModel(model: string): void;
    cleanup(): void;
}

/**
 * OpenRouter provider implementation for the AgentForce SDK
 * Handles communication with OpenRouter's API using OpenAI-compatible interface
 */
export class OpenRouterProvider implements OpenRouterProviderInterface {
    private model: string;
    private modelConfig?: ModelConfig;
    private client: OpenAI;
    private toolUse: OpenRouterToolUse;

    constructor(model: string, modelConfig?: ModelConfig) {
        this.model = model;
        this.modelConfig = modelConfig;
        
        const apiKey = process.env.OPENROUTER_API_KEY;
        if (!apiKey) {
            throw new Error("OPENROUTER_API_KEY environment variable is required");
        }

        this.client = new OpenAI({
            baseURL: "https://openrouter.ai/api/v1",
            apiKey: apiKey,
            defaultHeaders: {
                "HTTP-Referer": process.env.YOUR_SITE_URL || "https://agentforce.zone",
                "X-Title": process.env.YOUR_SITE_NAME || "AgentForce ADK",
            },
        });

        this.toolUse = new OpenRouterToolUse(model, modelConfig);
    }

    /**
     * Generate response using the OpenRouter model
     * @param prompt - The user prompt to send to the model
     * @param system - Optional system prompt to override the model's default
     * @returns Promise with the model's response
     */
    async generate(prompt: string, system?: string): Promise<string> {
        try {
            const messages: ChatCompletionMessageParam[] = [];
            
            if (system) {
                messages.push({ role: "system", content: system });
            }
            
            messages.push({ role: "user", content: prompt });

            const completion = await this.client.chat.completions.create({
                model: this.model,
                messages: messages,
                ...(this.modelConfig?.temperature !== undefined && { temperature: this.modelConfig.temperature }),
                ...(this.modelConfig?.maxTokens !== undefined && { max_tokens: this.modelConfig.maxTokens }),
            });

            return completion.choices[0]?.message?.content || "";
        } catch (error) {
            const errorMessage = error instanceof Error ? error.message : String(error);
            return `Error: OpenRouter provider error - ${errorMessage}`;
        }
    }

    /**
     * Chat with the OpenRouter model
     * @param messages - Array of messages for the conversation
     * @returns Promise with the model's response
     */
    async chat(messages: Array<{ role: string; content: string }>): Promise<string> {
        try {
            // Convert messages to OpenAI format
            const openAIMessages: ChatCompletionMessageParam[] = messages.map(msg => ({
                role: msg.role as "system" | "user" | "assistant",
                content: msg.content,
            }));

            const completion = await this.client.chat.completions.create({
                model: this.model,
                messages: openAIMessages,
                ...(this.modelConfig?.temperature !== undefined && { temperature: this.modelConfig.temperature }),
                ...(this.modelConfig?.maxTokens !== undefined && { max_tokens: this.modelConfig.maxTokens }),
            });

            return completion.choices[0]?.message?.content || "";
        } catch (error) {
            const errorMessage = error instanceof Error ? error.message : String(error);
            return `Error: OpenRouter provider error - ${errorMessage}`;
        }
    }

    /**
     * Get the current model name
     */
    getModel(): string {
        return this.model;
    }

    /**
     * Set a new model
     */
    setModel(model: string): void {
        this.model = model;
        this.toolUse = new OpenRouterToolUse(model, this.modelConfig);
    }

    /**
     * Generate response with tool support using the OpenRouter model
     * @param prompt - The user prompt to send to the model
     * @param tools - Array of tool definitions
     * @param system - Optional system prompt
     * @param logger - Optional logger for debugging
     * @param agent - Optional agent instance for MCP tool execution
     * @returns Promise with the model's response after tool execution
     */
    async generateWithTools(prompt: string, tools: Tool[], system?: string, logger?: AgentForceLogger, agent?: any): Promise<string> {
        return this.toolUse.generateWithTools(prompt, tools, system, logger, agent);
    }

    /**
     * Chat with tool support
     * @param messages - Array of messages for the conversation
     * @param tools - Array of tool definitions
     * @param logger - Optional logger for debugging
     * @param agent - Optional agent instance for MCP tool execution
     * @returns Promise with the model's response after tool execution
     */
    async chatWithTools(
        messages: Array<{ role: string; content: string }>,
        tools: Tool[],
        logger?: AgentForceLogger,
        agent?: any,
    ): Promise<string> {
        return this.toolUse.chatWithTools(messages, tools, logger, agent);
    }

    /**
     * Cleanup HTTP connections to allow process to exit cleanly
     */
    cleanup(): void {
        // Force close any HTTP connections
        if (this.client && (this.client as any)._httpAgent) {
            const agent = (this.client as any)._httpAgent;
            if (agent && typeof agent.destroy === "function") {
                agent.destroy();
            }
        }
    }
}