Agent guardrails and input validation

★ 7.6 · security

add-guardrails is a Claude Code skill that adds protective mechanisms and validation to the Deus personal AI assistant, preventing undesired agent behavior during task execution. It addresses issues like infinite tool-call loops and uncontrolled agent actions — critical safeguards given that Deus runs background agents for tasks such as automated documentation updates and scheduled summaries. Because Deus operates entirely in local Docker containers with no cloud sync, these guardrails help maintain session coherence and predictable behavior across long interactions. The skill is aimed at developers configuring Deus for automation workflows who need reliable, bounded agent execution.