RunwayML + Luma AI MCP Server

import { z } from 'zod'; import RunwayML from '@runwayml/sdk'; import LumaAI from 'lumaai'; // Import Luma AI SDK import axios from 'axios'; import { McpError, ErrorCode } from '@modelcontextprotocol/sdk/types.js'; import { pollRunwayTask } from '../utils/polling.js'; import { pollLumaTask } from '../utils/luma-polling.js'; // Revert back to .js extension // --- Zod Schema for Input Validation --- const providerEnum = z.enum(['runwayml', 'lumaai']).optional().default('runwayml'); // Luma specific enums/types based on their docs const lumaAspectRatios = z .enum(['16:9', '1:1', '3:4', '4:3', '9:16', '9:21', '21:9']) .optional() .default('16:9'); const lumaModels = z .enum(['ray-flash-2', 'ray-2', 'ray-1-6']) .optional() .default('ray-2'); // Default to Ray 2 as per docs // Runway specific enums/types (Updated for 2024-11-06 API version) const runwayResolutions = z.enum(['1280:768', '768:1280']).optional(); export const generateTextToVideoSchema = z.object({ provider: providerEnum, promptText: z.string().min(1, 'Prompt text cannot be empty.'), // Runway specific (optional) runway_model: z.string().optional(), // e.g., "gen-2" runway_resolution: runwayResolutions, // Use resolution instead of ratio runway_watermark: z.boolean().optional().default(false), // Luma specific (optional) luma_model: lumaModels, luma_aspect_ratio: lumaAspectRatios, luma_loop: z.boolean().optional(), // Luma supports loop // Common (optional) - use provider defaults if not specified duration: z.number().int().positive().optional(), // Luma uses string like "5s", Runway uses number seed: z.number().int().optional(), }); // Type alias for inferred schema type type TextToVideoParams = z.infer<typeof generateTextToVideoSchema>; // --- Tool Implementation --- export async function generateTextToVideoTool(args: any, exchange: any) { let validatedArgs: TextToVideoParams; try { validatedArgs = generateTextToVideoSchema.parse(args); } catch (error: any) { console.error( `[${new Date().toISOString()}] Invalid arguments for generate_text_to_video:`, error.errors ); throw new McpError( ErrorCode.InvalidParams, `Invalid arguments: ${error.errors .map((e: any) => `${e.path.join('.')} - ${e.message}`) .join(', ')}` ); } const { provider, promptText, runway_model = 'gen-2', // Default runway model runway_resolution, // Use resolution runway_watermark, luma_model, // Default handled by Zod luma_aspect_ratio, // Default handled by Zod luma_loop, duration, // Common param seed, // Common param } = validatedArgs; const progressToken = exchange?.progressToken; // Get progress token if available let taskId: string; let providerName: string = provider; // To use in messages try { console.log( `[${new Date().toISOString()}] Starting text-to-video generation via ${providerName} with prompt: "${promptText.substring( 0, 50 )}..."` ); /* // Commenting out non-functional RunwayML text-to-video block if (provider === 'runwayml') { // --- RunwayML Logic --- const apiKey = process.env.RUNWAYML_API_SECRET; if (!apiKey) { throw new McpError( ErrorCode.InternalError, 'RUNWAYML_API_SECRET is not configured on the server.' ); } const runwayClient = new RunwayML({ apiKey }); const endpoint = 'https://api.runwayml.com/v1/text_to_video'; // Use production endpoint const headers = { Authorization: `Bearer ${apiKey}`, 'Content-Type': 'application/json', }; // Note: Runway duration is number (seconds), Luma is string (e.g., "5s") const payload: Record<string, any> = { prompt: promptText, // Use 'prompt' key for RunwayML API model: runway_model, duration, // Runway uses number directly seed, watermark: runway_watermark, resolution: runway_resolution, // Use resolution key }; Object.keys(payload).forEach( (key) => payload[key] === undefined && delete payload[key] ); const response = await axios.post(endpoint, payload, { headers }); taskId = response.data?.id; if (!taskId) { throw new McpError( ErrorCode.InternalError, 'RunwayML API did not return a task ID.' ); } console.log( `[${new Date().toISOString()}] RunwayML task initiated with ID: ${taskId}` ); // Start Runway polling pollRunwayTask({ runwayClient, taskId, mcpExchange: exchange, progressToken, }).catch((error: any) => { // Added type annotation for error console.error( `[${new Date().toISOString()}] RunwayML polling failed for task ${taskId}:`, error.message || error ); }); // --- End RunwayML Logic --- } */ // End of commented out block if (provider === 'lumaai') { // Changed 'else if' to 'if' // --- Luma AI Logic --- const apiKey = process.env.LUMAAI_API_KEY; if (!apiKey) { throw new McpError( ErrorCode.InternalError, 'LUMAAI_API_KEY is not configured on the server.' ); } const lumaClient = new LumaAI({ authToken: apiKey }); // Luma uses duration as string like "5s" const lumaDuration = duration ? `${duration}s` : undefined; // Build Luma payload carefully, only adding defined values const lumaPayload: LumaAI.GenerationCreateParams = { prompt: promptText, // Prompt is required }; // Conditionally add optional properties ONLY if they are defined if (luma_model !== undefined) { lumaPayload.model = luma_model; } if (luma_aspect_ratio !== undefined) { lumaPayload.aspect_ratio = luma_aspect_ratio; } if (luma_loop !== undefined) { lumaPayload.loop = luma_loop; } // Ensure lumaDuration is a string before assigning const finalLumaDuration = duration ? `${duration}s` : undefined; if (finalLumaDuration !== undefined) { lumaPayload.duration = finalLumaDuration; } // Add seed if Luma supports it and it's provided // if (seed !== undefined) lumaPayload.seed = seed; // Uncomment if Luma adds seed support const generation = await lumaClient.generations.create(lumaPayload); taskId = generation.id || ''; // Ensure taskId is a string if (!taskId) { throw new McpError( ErrorCode.InternalError, 'Luma AI API did not return a generation ID.' ); } console.log( `[${new Date().toISOString()}] Luma AI task initiated with ID: ${taskId}` ); // Start Luma polling pollLumaTask({ lumaClient, generationId: taskId, mcpExchange: exchange, progressToken, }).catch((error: any) => { // Added type annotation for error console.error( `[${new Date().toISOString()}] Luma AI polling failed for task ${taskId}:`, error.message || error ); }); // --- End Luma AI Logic --- } else { // Should not happen due to Zod validation, but good practice throw new McpError(ErrorCode.InvalidParams, `Invalid provider: ${provider}`); } // Send initial confirmation back to the client (common for both) exchange.sendProgress?.({ token: progressToken, value: { status: 'INITIATED', taskId, provider: providerName }, }); // Return immediately after initiating the task and polling return { content: [ { type: 'text', text: `${providerName} text-to-video task initiated with ID: ${taskId}. Generation is in progress. Status updates will follow.`, }, ], }; } catch (error: any) { console.error( `[${new Date().toISOString()}] Error initiating ${providerName} text-to-video task:`, error ); // Handle specific API errors if possible, otherwise general internal error if (provider === 'runwayml' && axios.isAxiosError(error)) { const status = error.response?.status; const message = error.response?.data?.message || error.message || 'Unknown Axios error'; let mcpErrorCode = ErrorCode.InternalError; if (status === 401 || status === 403) mcpErrorCode = ErrorCode.InvalidRequest; else if (status === 400 || status === 422) mcpErrorCode = ErrorCode.InvalidParams; throw new McpError( mcpErrorCode, `RunwayML API Error: ${message} (Status: ${status || 'N/A'})` ); } else if (provider === 'lumaai' && error instanceof LumaAI.APIError) { let mcpErrorCode = ErrorCode.InternalError; if (error.status === 401 || error.status === 403) mcpErrorCode = ErrorCode.InvalidRequest; else if (error.status === 400 || error.status === 422) mcpErrorCode = ErrorCode.InvalidParams; throw new McpError( mcpErrorCode, `Luma AI API Error: ${error.message} (Status: ${error.status})` ); } // General fallback error throw new McpError( ErrorCode.InternalError, `Failed to initiate ${providerName} text-to-video task: ${error.message}` ); } }