TL;DR: Hermes now lets you assign a separate auxiliary model to review recent agent work via a new /review workflow.
Summary: The Hermes AI agent tool introduced a new /review command with configurable auxiliary model support. Running /review takes the last 10 messages plus any appended prompt, spawns a subagent using that chosen model to review the presented work, and returns the review to the main agent. The feature aims to give developers a dedicated quality-check loop without disrupting the main agent's flow.
Why it matters: This lets AI builders integrate a second, potentially stronger or cheaper model as a critic in their agent pipelines, improving reliability through separation of generation and review. Try setting a high-accuracy model as the auxiliary reviewer to catch errors before final output.
Source: x_com