ai_comment_guard 1.0.0

This is the first alpha release of AI Comment Guard. It is functional and
ready for testing on non-production sites. Feedback on bugs, edge cases,
and API behaviour is very welcome via the issue queue.

What's included

• AI-powered comment moderation via the Anthropic Claude API
• Three moderation modes: Flag (unpublish for review), Replace (redact harmful phrases), Block (replace entire body)
• Configurable AI sensitivity: Low, Medium, High
• Custom rules field — extend the AI prompt with plain-language site policy
• Auto-unpublish flagged and blocked comments
• Bypass permission for trusted roles
• Optional email notification to comment authors
• Violation logging to Drupal watchdog
• Status Report warning when the API key is not configured
• Compatible with Drupal 10, 11, and 12

Known limitations in this release

• Only the default comment_body field is moderated — comments with custom body fields are not yet supported
• No automated test coverage yet — being added before beta
• Moderation runs synchronously on comment submit — high-latency API responses may slow the form submission
• Only supports Anthropic Claude as the AI provider — multi-provider support planned for a future release

Upgrading

This is the initial release. No upgrade path needed.