Symptom

A prebuilt agent keeps calling tools step after step and only stops when GraphRecursionError is raised.

Evidence what you'll see

Run ends with a recursion-limit error instead of a final answer
Trace shows the same tool called with the same or nearly the same arguments

Cause

The step cap only counts hops. Nothing in the graph notices that the agent is making no progress, so a repeated tool call runs until the cap.

Fix steps

0 of 5 done

Step 1Open the trace and confirm the repeated calls share one tool name and argument set.
Step 2Check the tool's output: an empty or unhelpful result often makes the agent retry the same call.
Step 3Make the tool return a clear, actionable message when it finds nothing, so the agent can change course.
Step 4Add a no-progress check: stop or route to a human after N identical tool calls in a row.
Step 5Keep recursion_limit as a hard backstop, set to a value that matches the task's real depth.

Your check marks stay in this browser only.

Verify the fix worked

Re-run the failing input: the agent finishes or stops on the no-progress rule well before the cap.
The trace shows no more than N identical consecutive tool calls.

Did this fix it?

Sources and credit

Summarized in our own words from these public reports. Credit goes to their authors.

langgraph issue #6731 ↗

Provided as is, without warranty. Test in a non-production environment and back up first. Security-related steps: verify against the official docs. Terms · Request removal or attribution change