Your agent keeps calling the same tool with the same arguments in a loop. How do you debug and fix it?

30-second answerSay your answer out loud first, then reveal.

Step 1 — Diagnose from traces

Look at what the tool returned. Common root causes:

Symptom in trace	Likely cause	Fix
Tool returns empty list / "no results"	Model thinks retrying will help	Return guidance: "No orders found for this email. Ask the user to confirm the email."
Tool returns cryptic error (500, stack trace)	Model can't self-correct	Return actionable errors
Result is truncated without saying so	Model keeps searching for the "rest"	Say "truncated; showing 10 of 300" and add pagination
Result is correct but model ignores it	Instructions unclear about what "done" means, or result buried in noise	Clarify success criteria; slim down output
Two tools overlap	Model alternates between them	Merge or clarify descriptions

Fix it in the tool design or prompt (see the table).

Detect repetition: hash (tool_name, args); if it is seen N times (e.g. 3), don't execute. Inject a message: "You've already called X with these arguments and got Y. Try a different approach or report what you found."
Max iterations, and a budget per run.
Escalate to a human or return a partial answer when stuck.

Add this case to your eval set and track a "repeated tool call rate" metric.

Only increasing max_iterations or adding "DON'T LOOP" to the prompt. That treats the symptom, not the cause.

Slow is fine. Stopping is the only problem.