Skip to content

Oxyjen v0.2.0 - Memory-aware LLM Runtime Foundation

Choose a tag to compare

View all tags
@11divyansh 11divyansh released this 23 Jan 10:21
· 408 commits to main since this release
v0.2.0
703ac68

The execution layer of Oxyjen is here.

This release introduces memory-aware LLM nodes and deterministic execution semantics, laying the foundation for reliable AI orchestration in Java.

What’s in 0.2.0

  • LLMNode - Native graph node for LLM execution

  • Context-aware Memory - Ordered, persistent memory via NodeContext

  • LLMChain - Deterministic retry & fallback execution

  • Public LLM API - LLM.of(), LLM.profile(), LLM.chain()

  • Explicit error model - Clear retry vs fail-fast semantics

  • OpenAI transport - Clean provider integration with error classification

Design Focus

  • Convention over configuration

  • LLMs as first-class graph nodes (not utilities)

  • Separation of policy vs enforcement (timeouts are declarative in v0.2)

  • Fully testable execution behavior

Notes

  • Timeout enforcement, DAG execution, and concurrency are planned for v0.3

  • OpenAI smoke tests are manual only (no CI dependency on API keys)

Assets 2
Loading