ax

Ax generates JSON Schema from signatures for structured LLM output using declarative field definitions and validation constraints. Learn how Ax delivers precise structured data.

Unlock chain-of-thought reasoning in Ax LLM. Configure thinking token budgets for enhanced AI responses. Supports OpenAI, Gemini, and Anthropic.

Learn how Ax enables provider-agnostic context caching for LLMs. Discover how to use the .cache() modifier to serialize reusable prompt segments for efficient LLM interactions.

Easily configure rate limiting for API calls made by Ax. Ax's token-based rate limiter tracks usage and delays requests to manage your token budget effectively.

Update Ax projects from v13 to v15 seamlessly. Learn strategies to replace factory functions, adopt new classes, and ensure type safety for future compatibility.

Debug Ax applications and trace AI operations with OpenTelemetry. Configure the Node SDK to monitor prompts to responses and control payload visibility for effective troubleshooting.

Learn to write unit and integration tests for Ax code using Vitest patterns. Discover deterministic LLM mocks with AxMockAIService for efficient testing, reducing API dependencies.

Secure your Ax applications by configuring AxJSRuntime for robust JavaScript sandboxing. Block dangerous globals and grant selective permissions to enhance security.

Implement a custom LLM provider for Ax by extending AxAIService and wrapping it with AxAIProvider. Integrate any language model seamlessly with the Ax SDK for full compatibility. Get the complete guide.

Master Ax memory for context management. Learn to track interactions, rewind with tags, and integrate persistence using AxMemory and MCP servers.

Learn to handle multi-modal inputs like images and audio in Ax applications. Discover how AxChatRequest processes various media or uses fallback strategies for seamless integration.

Learn how Ax implements streaming output with an asynchronous generator for real-time LLM response consumption. Explore the streamingForward() method and handle incremental deltas efficiently.