content-policy

MACE (MCP-Assisted Content Enforcement) is an MCP server for Adobe Experience Manager (AEM) that scans and fixes regional style governance violations using native AEM APIs—no custom AEM code. Works with Claude Desktop and Cursor.

Independent YouTube evaluation framework for content policy classification. Dual-classifier architecture (LLM + rules baseline) across YouTube's 14 advertiser-friendly categories with cross-classifier analysis, consistency auditing, and a Streamlit dashboard.

WordPress content moderation and profanity filtering plugin (development preview) for posts, pages and comments. Use Unicode-aware term matching, Persian/Arabic normalization, versioned policy rules, shadow mode, pending-review enforcement, a moderation playground and privacy-focused audit logs. Not production-ready.

The VEIL Standard: an open, signed permission envelope for AI-assisted sessions. Eight content axes with ceilings and floors, verification tiers, and a Guardian tier giving parents a verifiable way to consent to a young person's supervised AI use. CC BY-SA 4.0.

TayyibLens · Pure by design. Cited fiqh evidence for AI-generated images and audio.

Design a comprehensive prompt-based system for content moderation that handles nuanced cases, cultural contexts, and edge cases, with explainable decisions and adjustable sensitivity levels.

Benchmark Suite for Local Models — character, coding, agent, vision, and content policy evaluation via LM Studio