Done Is Not the Same as True
A successful execution result and the durable truth of its effect are related, but they are not automatically the same proposition.
Read 01 →A technical view of Baranes for readers who want to inspect lifecycle, effect boundaries and evidence flow. Every view distinguishes current evidence, open-source code and target architecture.
The public Atlas starts with RepoOps Architecture: lifecycle, persistent artifacts, bounded candidate boundary, governed Git adoption, execution roles, governance primitives, identity trace and inspectable cognitive cards.
The diagram is primarily a target architecture cognitive model. SDD lifecycle contracts are separately open source; some RepoOps mechanisms are implemented and evidenced separately, but the map must not be read as a production-deployment claim.
The public Atlas is deliberately scoped to product architecture, effect boundaries and evidence flow.
AI systems can execute actions, observe outcomes, verify evidence and record facts. The difficult part is keeping those different acts from silently becoming the same authority.
This short series explains three boundaries in Baranes: the difference between finishing an action and establishing its effect, who owns that effect determination, and what strong evidence can — and cannot — prove.
A successful execution result and the durable truth of its effect are related, but they are not automatically the same proposition.
Read 01 →Baranes leaves authoritative effect determination with the semantic domain whose state was affected — without requiring one service per domain.
Read 02 →Hashes, signatures, provenance and attestations can strengthen evidence without turning the evidence layer into a truth authority.
Read 03 →Additional views appear only when they clarify the public product architecture without turning every source diagram into a product claim.