@inproceedings{wang-2026-triage,
title = "{TRIAGE}: Severity-Ranked Multi-Agent Failure Attribution",
author = "Wang, Xizhi",
editor = "Zhang, Yue and
Xu, Ruifeng and
Purwarianti, Ayu and
Rong, Tong and
Suzuki, Jun",
booktitle = "Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the {A}sia-Pacific Chapter of the {A}ssociation for {C}omputational {L}inguistics (Volume 2: Short Papers)",
month = nov,
year = "2026",
address = "Hengqin, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.aacl-short.27/",
pages = "376--397",
ISBN = "979-8-89176-440-8",
abstract = "Failure attribution for multi-agent LLM systems is typically formulated as identifying a single decisive error in a failed execution trace. Yet failed traces often contain multiple interacting errors and admit several plausible intervention points. To examine whether an annotated decisive error is also the most promising point for repair, we introduce TriageBench, comprising LLM-council judgments over 1,184 Who When intervention candidates. The council prioritizes a different step from the original label in 59.3{\%} of traces. We therefore reframe failure attribution as severity-ranked diagnostic triage and instantiate it as TRIAGE, a zero-shot, cost-aware cascade combining LLM-as-judge suspicion scoring, structural impact, and LLM-simulated replay over generated repair patches. Under top-$K$ evaluation, TRIAGE reaches 73.9{\%} top-5 on Who When (vs. 42.4{\%} top-1) and 93.2{\%} top-10 on multi-error TRAIL (vs. 56.8{\%} top-1). A separate evaluation of generated top-3 repair sets finds at least one plausible repair on 98.5{\%} / 99{\%} of 326 traces (human rater / 3-LLM panel) under lenient (good $+$ ok) rating. Our code (https://github.com/jimmywang585/triage-failure-attribution) and TriageBench (https://huggingface.co/datasets/jimmywang585/triage-bench) are publicly available."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-2026-triage">
<titleInfo>
<title>TRIAGE: Severity-Ranked Multi-Agent Failure Attribution</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xizhi</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 2: Short Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruifeng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ayu</namePart>
<namePart type="family">Purwarianti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tong</namePart>
<namePart type="family">Rong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Suzuki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hengqin, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-440-8</identifier>
</relatedItem>
<abstract>Failure attribution for multi-agent LLM systems is typically formulated as identifying a single decisive error in a failed execution trace. Yet failed traces often contain multiple interacting errors and admit several plausible intervention points. To examine whether an annotated decisive error is also the most promising point for repair, we introduce TriageBench, comprising LLM-council judgments over 1,184 Who When intervention candidates. The council prioritizes a different step from the original label in 59.3% of traces. We therefore reframe failure attribution as severity-ranked diagnostic triage and instantiate it as TRIAGE, a zero-shot, cost-aware cascade combining LLM-as-judge suspicion scoring, structural impact, and LLM-simulated replay over generated repair patches. Under top-K evaluation, TRIAGE reaches 73.9% top-5 on Who When (vs. 42.4% top-1) and 93.2% top-10 on multi-error TRAIL (vs. 56.8% top-1). A separate evaluation of generated top-3 repair sets finds at least one plausible repair on 98.5% / 99% of 326 traces (human rater / 3-LLM panel) under lenient (good + ok) rating. Our code (https://github.com/jimmywang585/triage-failure-attribution) and TriageBench (https://huggingface.co/datasets/jimmywang585/triage-bench) are publicly available.</abstract>
<identifier type="citekey">wang-2026-triage</identifier>
<location>
<url>https://aclanthology.org/2026.aacl-short.27/</url>
</location>
<part>
<date>2026-11</date>
<extent unit="page">
<start>376</start>
<end>397</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TRIAGE: Severity-Ranked Multi-Agent Failure Attribution
%A Wang, Xizhi
%Y Zhang, Yue
%Y Xu, Ruifeng
%Y Purwarianti, Ayu
%Y Rong, Tong
%Y Suzuki, Jun
%S Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 2: Short Papers)
%D 2026
%8 November
%I Association for Computational Linguistics
%C Hengqin, China
%@ 979-8-89176-440-8
%F wang-2026-triage
%X Failure attribution for multi-agent LLM systems is typically formulated as identifying a single decisive error in a failed execution trace. Yet failed traces often contain multiple interacting errors and admit several plausible intervention points. To examine whether an annotated decisive error is also the most promising point for repair, we introduce TriageBench, comprising LLM-council judgments over 1,184 Who When intervention candidates. The council prioritizes a different step from the original label in 59.3% of traces. We therefore reframe failure attribution as severity-ranked diagnostic triage and instantiate it as TRIAGE, a zero-shot, cost-aware cascade combining LLM-as-judge suspicion scoring, structural impact, and LLM-simulated replay over generated repair patches. Under top-K evaluation, TRIAGE reaches 73.9% top-5 on Who When (vs. 42.4% top-1) and 93.2% top-10 on multi-error TRAIL (vs. 56.8% top-1). A separate evaluation of generated top-3 repair sets finds at least one plausible repair on 98.5% / 99% of 326 traces (human rater / 3-LLM panel) under lenient (good + ok) rating. Our code (https://github.com/jimmywang585/triage-failure-attribution) and TriageBench (https://huggingface.co/datasets/jimmywang585/triage-bench) are publicly available.
%U https://aclanthology.org/2026.aacl-short.27/
%P 376-397
Markdown (Informal)
[TRIAGE: Severity-Ranked Multi-Agent Failure Attribution](https://aclanthology.org/2026.aacl-short.27/) (Wang, AACL-IJCNLP 2026)
ACL
- Xizhi Wang. 2026. TRIAGE: Severity-Ranked Multi-Agent Failure Attribution. In Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 2: Short Papers), pages 376–397, Hengqin, China. Association for Computational Linguistics.