AxLLM

AxLLM is a powerful, production-ready TypeScript AI framework designed to simplify building complex systems and AI workflows using large language models (LLMs). It provides a standardized, type-safe interface across top LLM providers, automatically generating efficient prompts from declarative signatures called "prompt signatures." AxLLM supports advanced capabilities such as end-to-end streaming, multi-agent work orchestration, multi-modal input/output (text, images, audio), retrieval-augmented generation (RAG), and function/tool calling aligned with the Model Context Protocol (MCP).

Key features include:

• Type-safe prompt signatures: Define input/output types and descriptions in a concise DSL to auto-generate optimized prompts with strong TypeScript integration.
• Multi-model orchestration (AxFlow): Build complex AI workflows with automatic parallel execution, adaptive control flow, multi-agent coordination, loops, branching, error handling, and streaming support.
• Multi-provider API abstraction: Integrates with OpenAI, Google Gemini, Anthropic, Cohere, HuggingFace, and others, enabling model load balancing and routing with type-safe model selection.
• Native support for multi-modal AI: Including images and audio as input or output fields supporting models like GPT-4o and Gemini.

Use cases:

• Building autonomous AI agents that orchestrate multi-step reasoning and external tool usage.
• Creating complex AI pipelines that combine summarization, classification, generation, and retrieval.
• Managing multi-modal conversational assistants with streaming responses.