Consider adding support for Project CodeGuard rules in Guardrails.
Developers and researchers using the Guardrails functionality to build applications and agents that could generate code could benefit from Project CodeGuard's rules .
We could integrate Project CodeGuard's rules and methodology into the guardrails. This could involve an integration and/or a well-documented process for users to import and use the rules.
Allow the guardrail to run the CodeGuard analysis on code snippets generated by the model. The guardrail would be able to output a detailed report of any security issues found, trip a "blocker" to halt execution, or completely avoid introducing a new vulnerability.