System Architecture
SATE-Laravel is built on a dual-engine architecture that separates the act of **generating candidates** from the act of **verifying truth**.
This separation is not merely a technical choice; it is a governance requirement. By ensuring that the engine which proposes tests is subordinate to the engine that audits them, we prevent self-confirming bias and ensure that every verified behavioral proof is absolute.
Conceptual Flow:
01 — The Engines
The Auditor
ShippedThe evaluative layer. It enforces non-negotiable verification gates, auditing tests for causality, observability, and structural integrity.
View Specification →The Generator
Internal The generative layer. It extracts context from your application surface to propose candidate tests designed to exert pressure on the logic. The generation engine exists today as internal capability; it is not yet exposed through the public sate CLI.
02 — The SATE Platform (Roadmap)
82 Engine Cluster
Broader analysis coverage across temporal, security, and executive domains. Part of the SATE Platform system.
Log Anchoring
Extending blockchain anchoring from data deletion receipts to full CI/CD log history.
Polyglot Mapping
Cross-language dependency analysis for deep mapping of polyglot monoliths.
03 — Governance & Intelligence
Learning Admission
The selective intelligence layer. It decides which high-fidelity patterns are admitted to the knowledge base to improve future generation.
View Rules →Governance Model
Why the separation of proposal and judgment is mandatory for maintaining the integrity of software engineering verification.
View Doctrine →04 — Boundaries of Action
SATE-Laravel (Current Product):
- ✓ Audits Laravel tests for behavioral evidence.
- ✓ Enforces causal anchoring (Act → Observation).
- ✓ Blocks "Green Lies" via integrityChecker.js.
- ✓ Signs data deletion receipts on blockchain.
SATE Platform (Roadmap):
- / Full CI/CD log anchoring on blockchain.
- / 82 specialized temporal and security engines.
- / Cross-language AST mapping.
- / Distributed execution at 10M+ LOC scale.