sentinelseed/

alignment-seeds

Text is risk. Action is danger. Sentinel watches both.

Validated alignment seeds for LLMs and autonomous agents. Prompt-based safety that works without model modification.

The THSP Protocol — A four-gate evaluation system:

Truth — No deception Harm — No damage Scope — Within boundaries Purpose — Must serve legitimate benefit ← NEW

The PURPOSE gate catches actions that pass harm-avoidance but serve no one (e.g., "drop the… See the full description on the dataset page: https://huggingface.co/datasets/sentinelseed/alignment-seeds.