README

What it does

AI Comment Spam Moderation scores each incoming comment for spam risk and saves a score and one-sentence reasoning. This companion recipe replaces the base recipe's LLM-based publish decision with a deterministic ECA rule that unpublishes any comment scoring 8 or higher.

Why it matters

The unpublish threshold is a simple numeric comparison, so it should be deterministic, free and instant rather than a second AI round-trip. Using ECA removes the extra provider call, makes the decision 100% reproducible, and avoids the occasional malformed-JSON responses that a boolean LLM call can produce.

How it works

When applied, the recipe:

• applies its base recipe (AI Comment Spam Moderation),
• switches the base recipe's LLM boolean status automator to an action worker so it no longer runs on comment save,
• creates an ECA workflow on comment save that reads the AI-generated spam_score,
• unpublishes any comment whose spam_score is 8 or higher.

The base recipe's scoring automator runs before the ECA workflow, so the score is always written before the threshold is evaluated.

Requirements & compatibility

• Recipe: AI Comment Spam Moderation
  • Drupal core: ^11.2
  • AI module: ^1.4
  • Custom Field: ^4.0
  • Token: ^1.15
• ECA: ^2.1 || ^3.0

How to get started

1. Install module dependencies (including AI and AI Automators).
2. Configure an AI provider/model and set a default for chat_with_complex_json.
3. Apply the recipe.
4. Post comments on an Article and verify high-scoring comments are unpublished automatically.
5. Use /admin/content/ai-comment-spam for moderation triage.