nbaccess-separation-check

To find and fix NBAccess separation violations in Wolfram Language packages, use a two-phase scanning workflow that combines static pattern detection with LLM-based contextual judgment to identify access-control breaches. It then applies automated repairs via cached results to correct sensitive cell and FE state violations efficiently.

The NBAccess separation principle in Wolfram Language notebooks is a compliance rule requiring packages to access FE state and sensitive cells exclusively through public NBAccess APIs. It prevents direct access-control breaches by enforcing ClaudeCheckSeparation scanning workflows.

Yes, you can configure separation checks to ignore specific notebook patterns by defining the $NBSeparationIgnoreList variable before running the violation detection workflow. This allows you to set safety boundaries and skip compliant cells during the scanning process.

To verify NBAccess separation fixes using a specific model, set the $ClaudeTestModel variable to your desired objective model before executing the validation phase. This drives the LLM-based verification constraints to ensure applied repairs meet compliance standards.

Yes, the separation check workflow supports automated repair for ClaudeEval and ClaudeQuery codebases by running ClaudeFixSeparation to apply corrections. It uses cached findings from the initial detection phase to efficiently fix access-control violations without rescanning.

Static pattern scanning might miss Wolfram Language notebook access-control breaches because nuanced violations require contextual understanding of FE state interactions. Combining static scanning with LLM-based judgment in a two-phase detection process reduces false negatives and catches complex separation principle violations.