The advanced semantic and vector processing core. ScopeLogic synthesizes multi-model queries and constructs mathematically deterministic DAG proposals, definitively eliminating AI hallucinations before they reach execution.
ScopeLogic acts as a high-intelligence sandbox. It evaluates intent and generates logic, but operates under a strict architectural mandate: it possesses zero authority to persist, execute, or modify OS state without explicit human validation.
Runs bounded clarification questions. Enforces a strict one-question-per-round rule to accurately consolidate requirements without hallucinating user intent.
Transforms intent into structured JSON Directed Acyclic Graphs (DAGs). Hard-blocked from generating arbitrary executable bash, sh, or python code blocks.
Runs structural integrity (SI) and safety assertion (SA) checks independently of the model. Emits validation metadata and confidence scores to PortGrid.
Cross-references proposed logic against the verifiable BaseVault schema. Instantly rejects unknown node types, unmapped tools, and cyclical dependencies.