The Problem: I’m performing an action using the tool and interrupt. Once the tool executes certain process, the graph should interrupt. Because of the execution of processes, that produce different result on every execution, I can’t place the interrupt inside the tool, so I’m using the flag to redir…

Hi @neel , Short version — this is almost certainly a graph topology + Command(goto=…) interaction issue, not an IDE debugging artifact. Root Cause (Most Likely) You currently have: Static edge: interrupt_node → agent_node Tools returning: Command(goto="agent_node") And earlier: tool_nod…

hi @neel In LangGraph, interrupt(value) does not pause at that line and then continue from the next line later. Instead: The first call raises a GraphInterrupt and stops execution. When you resume with Command(resume=...), the value you pass becomes the return value of interrupt(), and the node …

Hi @pawel-twardziak , Thanks for the help. [image] pawel-twardziak: In LangGraph, interrupt(value) does not pause at that line and then continue from the next line later. Yeah, it is already mentioned in the docs , and that is the reason why I’m interrupting the graph in a node which has ide…

hi @neel thanks for following up. Two more questions: Do any of your tools (inside ToolNode) ever return Command(goto=…)? (Even in some branches) Besides START->agent_node and interrupt_node, do you also have a static edges out of agent_node or tool_node (in addition to returning Command(goto=…)…

Hi @pawel-twardziak All the tools are returning Command(goto=”agent_node”, update=…). I had only 2 edges at first, (START → agent_node), (tool_node →[ agent_node, interrupt_node]). The current version has 3rd edge (interrupt_node → agent_node) as suggested by Chat LangChain bot. That’s it, nothing…

Hi @pawel-twardziak , In addition to this problem, I have faced another situation as well, not frequently but it had occurred sometimes. So, I’m using the interrupt to process the complex user actions, and the result of that action is appended in the original tool call. The agent only sees the ric…

Hi @neel thanks, I’ll go through this again today

Yeah, it might be possible. Will let you know once I fix them. Thanks for the help, @Bitcot_Kaushal .

Hi @Bitcot_Kaushal & @pawel-twardziak I’ve replaced other routing methods with Command, and fixed the issue of double routing as well. Now, I get the expected behaviour from the agent/graph. Thanks for the help.

hi @neel great to hear you managed that! Wrapping up for the others: The final issue was double routing back to agent_node, which caused duplicate AI outputs after resume. In practice, routing was being mixed across mechanisms: static edges (for example interrupt_node -> agent_node), and Comm…

Hey @neel That’s great to hear! :blush: I’m really glad switching fully to Command-based routing resolved the issue and the graph is now behaving as expected. These routing conflicts can be subtle, but once cleaned up, things usually fall into place nicely. Thanks for sharing the update — and wel…

Twice execution of agent when using the interrupt

OSS Product Help LangGraph

pawel-twardziak March 2, 2026, 6:25am 15

thanks @neel

Topic		Replies	Views
Getting Multiple Interrupts at the same time from graph LangGraph python-help	1	373	October 23, 2025
Problems encountered when study is interrupted LangGraph intro-to-langgraph , python-help	1	42	March 7, 2026
Auto resuming challenges in langgraph LangGraph self-hosted , intro-to-langgraph , python-help	4	883	September 29, 2025
Having issue while using interrupt with Agent Chat UI Deployment	7	81	March 20, 2026
[LangSmith Studio Issue] when resuming from an interrupt inside a subgraph. it doesn't properly resume, instead restarts LangSmith Product Help python-help	10	126	March 19, 2026

Twice execution of agent when using the interrupt

Related topics