The Principles
How we act.
If the manifesto says what we value, the principles say how we act. Each one is born from a value.
The principles
No code without an approved specification
Every change begins with a specification a human approves before a line is written. Understanding precedes building.
from: Specification before implementation
The machine proposes, the human decides
The AI agent generates, suggests, and executes — but the decisions that matter pass through human judgment. Final authority is never delegated.
from: Human judgment at the decision points
Quality is validated, not assumed
Functionality and integrity are verified by independent roles who did not write the code. Trusting output without checking it isn't trust — it's negligence.
from: Independent, multi-role validation
Every decision leaves a trace
Who did it, when, and on what basis. What the AI generated, and how much was reviewed. A system no one can audit is a system out of control.
from: A traceable path for every decision
Clear roles, clear boundaries
Every participant — human or agent — knows what they may and may not do. Permission is explicit and enforced, not a social convention.
from: Human judgment at the decision points
The flow is visible to everyone
The state of every piece of work is transparent in real time. No one has to ask where something stands — the system shows it.
from: A traceable path for every decision
See how the principles play out in the work.