Skip to content
Discussion options

You must be logged in to vote

AARM — Agent Activity and Reasoning Monitoring — is the approach of monitoring both what an agent does (activity: tool calls, outputs) and why (reasoning traces), instead of one or the other. Combining behavioural and reasoning analysis improves detection of misaligned behaviour over activity-only monitoring. Adrian is an AARM-aligned implementation.

Framework: https://aarm.dev · Background: https://secureagentics.ai/blog-posts/aarm

Replies: 1 comment

Comment options

You must be logged in to vote
0 replies
Answer selected by gladstomych-sa
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Category
Q&A
Labels
None yet
1 participant