drupal-eca-recipe/eca_lib_0051

Scores every newly submitted comment for toxicity and spam with one AI call that returns a strict JSON verdict, then routes that verdict through ordinary ECA conditions into one of three exits: publish, hold for review, or block. The thresholds are two "Token: set value" actions in the model, not numbers in the prompt, so the policy can be changed without touching the prompt. Every run writes both scores, both thresholds and the exit taken to the eca_comment_verdict log channel, and a failed or undecodable AI response holds the comment instead of publishing it. Requires the ai_integration_eca and eca_tamper contributed modules, and a chat model that supports structured output has to be selected in the "Ask AI to score the comment" action.

Latest 1.0.0Packagist

Vulnerabilities

We didn't find any vulnerabilities for this package.

Direct Dependencies

Dependency	Last Release	Score
drupal/eca Version *	—	—
drupal/eca_tamper Version *	—	—
drupal/modeler_api Version *	—	—
drupal/ai_integration_eca Version *	—	—
drupal-eca-recipe/article_content_type Version *	—	—