echopath-labs/openDomain

What it solves

OpenDomain provides a way to maintain a "semantic layer" for business logic—such as rules, concepts, and lifecycles—that is readable by both humans and AI agents. It prevents AI agents from hallucinating or automatically altering business rules by requiring that any AI-inferred knowledge be treated as a "Candidate" and explicitly reviewed and accepted by a human before becoming part of the official domain knowledge.

How it works

The system stores business semantics in Git-native Markdown files with YAML front matter. It integrates with AI agents (specifically Codex) via a set of managed instructions and "Skills" that allow the agent to explore, model, and review domain knowledge. The workflow ensures a strict boundary: agents can propose changes (Candidates), but only humans can finalize the acceptance of those changes into the accepted domain knowledge.

Who it’s for

It is designed for software maintainers and AI agents working together on complex projects where business rules must be strictly governed and evidence-backed, ensuring that the AI's contributions to the domain model are verified.

Highlights

• Human-in-the-loop governance: AI-inferred knowledge must be human-reviewed as a Candidate before acceptance.
• Git-native storage: Uses Markdown and YAML for a transparent, version-controlled source of truth.
• Agent integration: Provides specific skills and contracts for AI agents to interact with the domain layer.
• Validation tools: Includes schema validation, reference integrity checks, and dependency graph diagnostics for multi-product workspaces.