@verydia/safety-agents
v0.0.1
Published
Safety evaluation agents for content moderation and harm detection.
Readme
@verydia/safety-agents
Safety evaluation agents for content moderation and harm detection.
Overview
@verydia/safety-agents provides agent-based safety evaluation for LLM inputs and outputs.
Features
- Safety Evaluator Agent: Content safety scoring
- Harm Detection: Multi-category harm detection
- Verdict Generation: Accept/reject/warning verdicts
- Integration: Works with agent-shell and workflow-runtime
Installation
pnpm add @verydia/safety-agentsOSS Boundary & Dependencies
This package is OSS-safe and production-ready.
Allowed Dependencies
@verydia/agent-shell- Agent execution framework@verydia/safety-core- Safety types and utilities@verydia/schema-core- Schema validation
All dependencies are strictly OSS-safe with no proprietary dependencies.
License
MIT
