Skip to main content
Oxygen (“oxy”) is an open-source, full-stack Data + AI platform for agentic data analytics. We combine a data modeling engine (ontology engine), an agent and automation engine, an agentic application engine, a data lakehouse, and an ETL engine, to become a one-stop shop for anything Data and AI. Oxygen is grounded in deterministic execution principles. Written in Rust and declarative by design, Oxygen provides the foundational components needed to transform AI-driven data analysis into reliable, production-ready systems through structured primitives, semantic understanding, and predictable execution. Oxygen applies software development lifecycle principles (e.g. build-test-deploy pipeline) to AI-driven data analytics. Oxygen establishes a structured workflow for data agents, involving agent creation, prompt testing, and production deployment. banner Specifically, this process is implemented as follows:
  • Agent Configuration: Define agents using .agent.yml files that specify their instructions. Provide context through .sql and .schema.yml files that supply queries and semantic information, respectively.
  • Workflow Development: Create .workflow.yml files to orchestrate multi-step processes. Chain validated agents and tasks together into reusable, production-ready automations.
  • Testing Framework: Add test cases directly to .agent.yml or .workflow.yml files. Execute tests using the oxy test command.
You can run individual agents with oxy run path/to/agent_name.agent.yml "question" or execute complete workflows using oxy run path/to/workflow_name.workflow.yml.

​
Resources

​
Get started

If you’re new to Oxy start here to learn the essentials.

Quickstart

Basic Oxy commands