← Back to Catalog

Agent Non-termination

A problem in which an agent or system does not reach an appropriate completion or stopping condition and continues unnecessary iterations or actions.

Example

After finding an option that satisfies the requirements, an agent repeats the same searches and comparisons without ending the task.

Distinctions and cautions

Non-termination is not merely a task taking a long time. Long-running work may be appropriate when progress continues and defined limits or completion conditions are respected. Premature Termination is the opposite problem. Planning or Tool-use Failure may contribute to non-termination, but no single cause is assumed.

Terminology note

What counts as unnecessary continuation may vary with completion criteria, resource limits, task design, and the amount of progress considered acceptable.

Evidence

View evidence →

When Agents Do Not Stop: Uncovering Infinite Agentic Loops in LLM Agents

Title
When Agents Do Not Stop: Uncovering Infinite Agentic Loops in LLM Agents
Author / organization
Hou et al.
Year
2026
Source type / publication status
Primary arXiv preprint; PREPRINT
Supported concepts / claim context
Agent Non-termination.
Scope and limitations
Analysis of 6,549 repositories; this dedicated large-scale source was not peer-reviewed as of the evidence cutoff.
Access / version note
2026 preprint; no published substitute used.
Recorded official URL / DOI / identifier
Preprint reference: arXiv:2607.01641

Robust Tool Use via Fission-GRPO

Title
Robust Tool Use via Fission-GRPO
Author / organization
Zhang et al.
Year
2026
Source type / publication status
ACL peer-reviewed research; published
Supported concepts / claim context
Tool-use recovery; repeated invalid invocation.
Scope and limitations
Tested training methods and models; method-specific.
Access / version note
Published ACL.
Recorded official URL / DOI / identifier
DOI: 10.18653/v1/2026.acl-long.1880 (DOI resolver)