Hermes Agent follows Claude Code's/loop: Agent wakes up to work

Hermes Agent follows Claude Code/loop, and the user allows Agent to repeat the same task in the current session, for example, by following the CI, checking the deployment status, or changing the code repeatedly until the test passes. Each trigger is a full Agent execution, rereading the state of the moment. The user can specify how often it will be executed or give the frequency to Hermes. The self-adaptation model is defaulted on a minute-by-minute check, and if the results remain unchanged several times, it is reduced to 2, 4 and 8 minutes, with a maximum of 15 minutes, once the result changes, returning to one minute. The cycle can also set a number of times and stop conditions, and Agent can judge for itself that the task has been completed, or use/loop-until to check that the additional judge conditions are met. The default runs to 100 times are suspended to avoid burning token when left unattended. This function follows Claude Code/loop directly and is compatible with even/proactive alias。