This page contains historical documentation from the BOA/RCA/RP era. It is not the current authoritative specification of VCDesign. Please refer to the current Core specifications (core/core.yaml, core/policies.yaml, core/metrics.yaml) for authoritative guidance.
This concept has been superseded by the Core + Policies + Metrics model. Refer to specs/core/ for current definitions.
This document represents the conceptual lineage that led to the current model.
BOA Core
BOA Core is
a set of structures for handling decision, responsibility, and stopping
in systems involving AI without breaking them.
This repository is not about producing the correct answer.
It exists to correctly stop when a decision cannot be made.
This Is Not a Framework
BOA / RCA / RP are not a layered architecture or framework in the general sense.
- They are not layers for dividing processing
- They are not role assignments for organizing responsibilities
They are all checkpoints that appear along the flow of decision.
The Three Included Structures
- BOA (Boundary-Oriented Architecture) — Declares the boundary where decisions are allowed
- RCA (Responsibility Closure Agent) — Takes responsibility for and completes the decision
- RP (Resolution Protocol) — Explicitly chooses to proceed, stop, or return the decision
The important point is that these do not exist simultaneously.
Basic Flow of Decision (Summary)
- The system reaches BOA
- RCA performs an Assessment (evaluation of willingness to take responsibility)
- RP activates only if RCA indicates “Accepted”
- RP always returns one of the following
- Resolve (decide)
- Reject (discard)
- Return (return to human or upstream)
- RCA closes responsibility for the decision
Why BOA Comes First
BOA is not the first technology. It is the first declaration of boundary.
- How far AI judgment is allowed
- Where it must stop
- Whose responsibility it becomes
If these are ambiguous, neither RP nor RCA can carry meaning.
Roles and Questions
BOA / RCA / RP each operate in different dimensions— similar to the relationship of place, person, and rule.
BOA — Boundary-Oriented Architecture
Question: Where should judgment occur?
BOA separates the system by boundary and defines where AI may judge and where humans must judge.
RCA — Responsibility Closure Agent
Question: Do you take responsibility for this judgment?
RCA is the subject that judges within the boundary, expressing ACCEPTED / DENIED / UNKNOWN.
RP — Resolution Protocol
Question: Can this be recognized as a formal resolution?
RP verifies procedural legitimacy—signature, required trace, and explicit responsibility.
Failure, Stop, and Return
The failure BOA Core avoids most is
allowing a decision to proceed when it cannot be made.
Stop is normal here. If boundary conditions are not met, the decision never starts.
Return is a responsible action. It explicitly declares: “not now,” “not under this condition,” or “return to human/design.”
Silence is not allowed. Doing nothing or treating failure as success erases responsibility.
BOA Core is not a design to automate judgment,
but a design to prevent judgment from breaking.