Janus is a minimal governance layer for human–AI development systems.
It introduces three structural invariants:
• append-only institutional logs
• dual evidence model (E+ / E−)
• explicit human authority decisions
Together these allow hybrid human–AI systems to remain traceable, auditable, and deterministically reconstructible.
https://janusgovernance.org/paper/janus-foundational-paper-v1.0.pdf
janus-governance-core — conceptual kernel and foundational paper
janus-rfc — protocol specifications
janus-runtime — experimental runtime
janus-docs — extended documentation
janus-governance.github.io — public website