mcp-server

This capability allows developers to define functions using a schema that can be called across multiple model providers. It leverages a registry to manage these functions and their respective APIs, enabling seamless integration with various LLMs. The architecture supports dynamic resolution of function calls, which allows for flexibility in switching between different model providers without changing the core application logic.

This capability provides a structured way to manage context for interactions with LLMs, allowing for improved relevance and coherence in responses. It uses a context stack that maintains the state of previous interactions, enabling the server to provide context-aware responses based on user input history. This approach enhances user experience by making interactions feel more natural and conversational.

This capability enables the server to dynamically orchestrate API calls to various LLMs based on user-defined criteria or application logic. It uses a rule-based engine to determine which API to call and how to format requests, allowing for optimized performance and cost management. This orchestration is particularly useful for applications that need to balance load across multiple models.