Glossary
Agentic loop
How an agent works: the model plans a step, acts through a tool, reads what came back and decides the next step, again and again until the task is done. Anthropic describes agents as LLMs using tools based on environmental feedback in a loop.
Every turn of the loop reads new text, and any of it can carry a prompt injection, so a loop that runs longer has more chances to be steered. Anthropic recommends stopping conditions such as a maximum number of iterations, testing in a sandbox and guardrails, and warns of higher costs and compounding errors.