The AI TQL Assistant plugin was added while the first draft of the secure coding guidelines related to AI usages were written.

Some adjustments are needed to respect them:

• basic guardrailing should be in place
• features must only be available to authenticated users
• metrics must be collected to make possible to identify users abusing the features
• logs must be collected
• inputs should have a limited length to limit the input token cost
• attempt at limiting the output length should be made to limit the output token cost