https://github.com/gwbischof/outsource-mcp - Code and Instructions can be found here.
Your current AI assistant might not always be the best choice to solve a problem. Sometimes you need a different model that's better suited for the specific task at hand. outsource-mcp fixes this by letting you delegate tasks to the right AI for the job.
Consider these real scenarios where you need different providers:
Haiku Generation Example
A user wants OpenAI to write a haiku about MCPs. The system delegates this creative task through outsource_text("openai", model, prompt) and returns contextually relevant poetry about Model Context Protocol.
DALL-E Image Generation Example
Generated Image from DALL-E
A user requests an image of the New York Knicks basketball team. The system routes this to outsource_image("openai", "dall-e-3", prompt) with a detailed prompt and successfully generates the requested image. This showcases how tasks can be delegated to models optimized for specific capabilities - DALL-E excels at image generation while many other models cannot create images at all.
NYC Wellness Search Example
A user requests Gemini to find wellness businesses in NYC using Google Maps data. The system routes this to outsource_text("google", "gemini", prompt) and returns structured business data with addresses, services, and contact information.
These examples demonstrate core challenges in multi-provider AI integration: different models have different capabilities and data access. DALL-E can generate images while Claude cannot, Gemini has access to Google Maps data while other models do not, yet accessing these specialized capabilities requires separate integrations and complex orchestration logic.
Modern AI agents are remarkably capable and can handle most tasks competently on their own. However, there are specific situations where outsourcing to different models becomes valuable: