LoopRails
A guardrail and risk control platform for AI agents. It provides a risk grading and scoring mechanism based on reversibility, scope of impact, and stakes, and supports authorization control patterns such as Maker-Checker. The platform is intended to constrain agent actions according to their assessed risk level and to enforce approval workflows for sensitive operations.
Timeline (last 90 days)
In LoopRails' terminology, Maker-Checker is a named pattern and is the concrete form the Authorized guardrail takes on the highest-risk actions.
Using the LoopRails scorer to rate by reversibility, scope of impact and stakes, G3-level (irreversible, large scope of impact or high stakes) actions are recommended to use Maker-Checker.
Research on AI coding agents found that a plan approval step reduced attacks, but the human intervention success rate was still only 9% to 26%.
All Facts (3)
In LoopRails' terminology, Maker-Checker is a named pattern and is the concrete form the Authorized guardrail takes on the highest-risk actions.
50%UnverifiedUsing the LoopRails scorer to rate by reversibility, scope of impact and stakes, G3-level (irreversible, large scope of impact or high stakes) actions are recommended to use Maker-Checker.
50%UnverifiedResearch on AI coding agents found that a plan approval step reduced attacks, but the human intervention success rate was still only 9% to 26%.
50%