Adaptive Runtime Security Platform for Autonomous Agents (v8.0)
A production-ready, drop-in governance layer for AI coding agents (Claude Code, Cursor, Aider).
It combines sandbox isolation, runtime telemetry, and weighted risk scoring to contain
prompt injection and malicious agent behavior before changes reach the host repository.
This is a defense-in-depth control, not a guarantee against all prompt injection — see
"Scope & Limitations" below.
Changelog: v7.2 →… See the full description on the dataset page: https://huggingface.co/datasets/Sahek/lean-agent-governance.
