ai4X is an operating model for explicit, reusable, and host-independent agentic AI work—agentic cognition engineering (ACE).
The project is rebuilding from a deliberately small, greenfield baseline. Product capabilities will return as focused, verified vertical slices rather than as a wholesale migration of the previous prototype.
- Need-first: describe the outcome before selecting implementation.
- Host-independent: keep project and agent declarations separate from host adapters.
- Agentic-first: make context discoverable, bounded, and executable by AI agents.
- Human-approvable: make intent, effects, evidence, and decisions understandable to people.
- Self-applicable: evolve ai4X using the same operating model it provides.
.ai4x/— canonical Project Memory and project declarationssrc/— semantic product source: foundation, domains, interfaces, and resourcesdoc/— human-readable documentation and durable evidenceutil/— repository development and verification tooling.github/,.codex/, andAGENTS.md— thin host integration facades
Start with .ai4x/BEHAVIOR.md and .ai4x/CONTEXT.md.
See CONTRIBUTING.md, GOVERNANCE.md, and LICENSING.md.