<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shamiullah-2026-failing">
    <titleInfo>
        <title>Failing Agents Leave Traces: Telemetry Signatures for Multi-Tier Agent Evaluation</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Muthaheera</namePart>
        <namePart type="given">Yasmeena</namePart>
        <namePart type="given">Belgur</namePart>
        <namePart type="family">Shamiullah</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-10</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the Second Workshop for REsearch on Agent Language Models (REALM 2026)</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Ehsan</namePart>
            <namePart type="family">Kamalloo</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Nicolas</namePart>
            <namePart type="family">Gontier</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Nikolai</namePart>
            <namePart type="family">Rozanov</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Maryam</namePart>
            <namePart type="family">Hashemzadeh</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Akhil</namePart>
            <namePart type="family">Arora</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Budapest, Hungary</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-485-9</identifier>
    </relatedItem>
    <abstract>Agent evaluation today depends on per-trace LLM-judge inference or human review, too expensive to run on every trace; production systems fall back to sampling a fraction. We find that failing agents leave a detectable behavioral signature in standard observability telemetry: disproportionate effort relative to outcome. We formalize four behavioral failure signatures from agent telemetry, validated on 4,671 traces spanning five agent domains: Compensatory Effort (17–159% more effort on each domain’s primary signal, all five domains, p\ensuremath&lt;0.006, the dominant signature every- where tested); Output Inflation (+38% larger patches on SWE-rebench; direction and reliability vary by domain); Unfocused Orchestra- tion (+5.98–12.3% higher token entropy, four of five domains); and Correlation Decoupling (effort-signal correlations shift structurally between pass and fail traces, p\ensuremath&lt;0.001, three domains). A simple threshold check over these signals, with no judge, no label, and no model call, already catches 52.9–61.5% of failures. Across three human-labeled benchmarks, judges miss 6.5–18.1% of confirmed failures on the two directly comparable ones, and a trained telemetry classifier (AUROC 0.821) recovers 61.5–75.4% of each judge’s misses; combining judge and classifier cuts the miss rate from 18.1% to 4.4%, a 4.1× reduction against the weakest of the eight judges tested (GPT-4o-mini); stronger judges start with fewer misses, so the reduction is correspondingly smaller.</abstract>
    <identifier type="citekey">shamiullah-2026-failing</identifier>
    <location>
        <url>https://aclanthology.org/2026.realm-1.56/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>816</start>
            <end>831</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Failing Agents Leave Traces: Telemetry Signatures for Multi-Tier Agent Evaluation
%A Shamiullah, Muthaheera Yasmeena Belgur
%Y Kamalloo, Ehsan
%Y Gontier, Nicolas
%Y Rozanov, Nikolai
%Y Hashemzadeh, Maryam
%Y Arora, Akhil
%S Proceedings of the Second Workshop for REsearch on Agent Language Models (REALM 2026)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-485-9
%F shamiullah-2026-failing
%X Agent evaluation today depends on per-trace LLM-judge inference or human review, too expensive to run on every trace; production systems fall back to sampling a fraction. We find that failing agents leave a detectable behavioral signature in standard observability telemetry: disproportionate effort relative to outcome. We formalize four behavioral failure signatures from agent telemetry, validated on 4,671 traces spanning five agent domains: Compensatory Effort (17–159% more effort on each domain’s primary signal, all five domains, p\ensuremath<0.006, the dominant signature every- where tested); Output Inflation (+38% larger patches on SWE-rebench; direction and reliability vary by domain); Unfocused Orchestra- tion (+5.98–12.3% higher token entropy, four of five domains); and Correlation Decoupling (effort-signal correlations shift structurally between pass and fail traces, p\ensuremath<0.001, three domains). A simple threshold check over these signals, with no judge, no label, and no model call, already catches 52.9–61.5% of failures. Across three human-labeled benchmarks, judges miss 6.5–18.1% of confirmed failures on the two directly comparable ones, and a trained telemetry classifier (AUROC 0.821) recovers 61.5–75.4% of each judge’s misses; combining judge and classifier cuts the miss rate from 18.1% to 4.4%, a 4.1× reduction against the weakest of the eight judges tested (GPT-4o-mini); stronger judges start with fewer misses, so the reduction is correspondingly smaller.
%U https://aclanthology.org/2026.realm-1.56/
%P 816-831

Markdown (Informal)

[Failing Agents Leave Traces: Telemetry Signatures for Multi-Tier Agent Evaluation](https://aclanthology.org/2026.realm-1.56/) (Shamiullah, REALM 2026)

• Failing Agents Leave Traces: Telemetry Signatures for Multi-Tier Agent Evaluation (Shamiullah, REALM 2026)

ACL

• Muthaheera Yasmeena Belgur Shamiullah. 2026. Failing Agents Leave Traces: Telemetry Signatures for Multi-Tier Agent Evaluation. In Proceedings of the Second Workshop for REsearch on Agent Language Models (REALM 2026), pages 816–831, Budapest, Hungary. Association for Computational Linguistics.