Auditor vs Generator
The intentional separation of proposal and judgment is the foundation of verification integrity. SATE-Laravel distinguishes these acts to prevent the collapse of truth within the system.
01 — Two Different Roles
The Generator
Proposes Candidates
Explores the behavioral surface and generates hypotheses of behavior in the form of executable code.
The Auditor
Evaluates Evidence
Inspects the integrity of test logic and determines whether a test provides a valid behavioral proof.
"Proposal and judgment are fundamentally different acts."
02 — Why Separation Is Mandatory
Mixing the system that proposes a solution with the system that validates it creates a closed feedback loop that corrupts judgment.
When a system evaluates its own output, it naturally tends toward self-confirmation. Bias becomes invisible because the same assumptions used during generation are applied during verification.
“A system that validates its own output cannot detect its own failure.”
03 — Consequences of Mixed Roles
Standard Decay
Verification gates gradually weaken to accommodate the limitations of the generation engine, reducing the overall rigor of the suite.
Execution Optimization
The system prioritizes tests that pass easily over tests that provide deep behavioral proof, optimizing for green metrics instead of truth.
Credibility Loss
Audit results lose their value because the "proof" is no longer independent of the proposal. The knowledge base becomes a circular reflection of its own bias.
04 — Audit Supremacy
In the SATE-Laravel architecture, the Auditor has the final and absolute say.
The verification gates are static and non-negotiable. They do not adapt to the Generator's capabilities or the developer's convenience. Nothing—regardless of its origin—bypasses the audit layer.
“Audit authority is absolute, or it is meaningless.”
05 — Generator Subordination
The Generator is a subordinate tool designed to exert pressure. Its output is granted no special status or presumed trust.
REJECTION IS NORMAL
Generated tests are expected to fail. Rejection is the primary feedback mechanism that prevents unanchored code from polluting the suite.
ZERO DEFAULT TRUST
Generation earns no trust by default. Every candidate must earn its place by satisfying the same gates as a human-authored test.
“Generation earns no trust by default.”
06 — Enforcement Models
SATE-Laravel provides the structural separation, but it does not mandate how teams enforce it. The separation is a design truth, not a policy mandate.
Strict Separation
Teams may choose to treat the Auditor as an absolute blocker, where no generated code is committed until it achieves a "Gold Standard" audit status.
Advisory Separation
Others may use the Auditor as a signaling tool, using generation for rapid exploration while acknowledging the trust gaps identified by the audit.
07 — Who This Is For
This boundary exists for those who value architectural certainty over the illusion of progress.
If your priority is the rapid accumulation of lines of code or the speed of "green" CI builds, this separation will feel restrictive and unnecessary.
“If speed matters more than certainty, this boundary will feel restrictive.”