Reversibility is the ability to undo an automated change if it causes an unwanted result. In cybersecurity automation, this means any action taken by a script, workflow, or AI agent should be easy to roll back, such as restoring a configuration, reopening access, or reverting a policy update.
It matters because security tools often make changes to live systems. A fast fix is useful only if it can be safely undone when it breaks connectivity, blocks legitimate users, or creates a new weakness. Defenders use reversibility as a control for agentic remediation: they pre-validate changes, log every action, and keep rollback steps ready. In attacks, reversibility is the opposite of what operators want; defenders rely on it to limit blast radius and recover quickly after a mistaken or overbroad automated response.