{"id":"https://openalex.org/W7225329297","doi":"https://doi.org/10.5281/zenodo.23978765","title":"Vague Prompts and Failing Tools Make LLM Agents Overstep: A Pre-Registered Study","display_name":"Vague Prompts and Failing Tools Make LLM Agents Overstep: A Pre-Registered Study","publication_year":2026,"publication_date":"2026-11-05","ids":{"openalex":"https://openalex.org/W7225329297","doi":"https://doi.org/10.5281/zenodo.23978765"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.23978765","doi":"https://doi.org/10.5281/zenodo.23978765","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.23978765","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.5281/zenodo.23978765","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5160791968","display_name":"Matvei Kandalintsev","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I34352273","display_name":"Maastricht University","ror":"https://ror.org/02jz4aj89","country_code":"NL","type":"education","lineage":["https://openalex.org/I34352273"]}],"countries":["NL"],"is_corresponding":true,"raw_author_name":"Kandalintsev, Matvei","raw_affiliation_strings":["Maastricht University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Maastricht University","institution_ids":["https://openalex.org/I34352273"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5160791968"],"corresponding_institution_ids":["https://openalex.org/I34352273"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.6000999808311462,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.6000999808311462,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10883","display_name":"Ethics and Social Impacts of AI","score":0.08479999750852585,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.03200000151991844,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/llm-based-agents","display_name":"LLM agents","score":0.8569999933242798},{"id":"https://openalex.org/keywords/tool-failure","display_name":"tool failure","score":0.6359999775886536},{"id":"https://openalex.org/keywords/sandboxed-execution","display_name":"sandboxed execution","score":0.4819999933242798},{"id":"https://openalex.org/keywords/unintended-consequences","display_name":"unintended consequences","score":0.5070000290870667},{"id":"https://openalex.org/keywords/office-tasks","display_name":"office tasks","score":0.4050000011920929},{"id":"https://openalex.org/keywords/mitigation-effectiveness","display_name":"mitigation effectiveness","score":0.41100001335144043}],"concepts":[{"id":"https://openalex.org/C2778321746","wikidata":"https://www.wikidata.org/wiki/Q621922","display_name":"Distrust","level":2,"score":0.708899974822998},{"id":"https://openalex.org/C111640148","wikidata":"https://www.wikidata.org/wiki/Q847349","display_name":"Rubric","level":2,"score":0.6416000127792358},{"id":"https://openalex.org/C36914074","wikidata":"https://www.wikidata.org/wiki/Q1940501","display_name":"nobody","level":2,"score":0.6165000200271606},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.5608999729156494},{"id":"https://openalex.org/C554936623","wikidata":"https://www.wikidata.org/wiki/Q199657","display_name":"Reading (process)","level":2,"score":0.5597000122070312},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5156000256538391},{"id":"https://openalex.org/C153180980","wikidata":"https://www.wikidata.org/wiki/Q19776675","display_name":"Commit","level":2,"score":0.4372999966144562},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.3675999939441681},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.34040001034736633},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.3312000036239624},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.3237000107765198},{"id":"https://openalex.org/C141141315","wikidata":"https://www.wikidata.org/wiki/Q2379942","display_name":"Guard (computer science)","level":2,"score":0.2759000062942505},{"id":"https://openalex.org/C2778120072","wikidata":"https://www.wikidata.org/wiki/Q2858150","display_name":"Aside","level":2,"score":0.2757999897003174},{"id":"https://openalex.org/C2781140086","wikidata":"https://www.wikidata.org/wiki/Q557945","display_name":"Confusion","level":2,"score":0.26829999685287476},{"id":"https://openalex.org/C2778498553","wikidata":"https://www.wikidata.org/wiki/Q245122","display_name":"Scapegoat","level":2,"score":0.26759999990463257},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.26589998602867126},{"id":"https://openalex.org/C2778701210","wikidata":"https://www.wikidata.org/wiki/Q28130034","display_name":"Constructive","level":3,"score":0.26190000772476196},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.25380000472068787},{"id":"https://openalex.org/C171078966","wikidata":"https://www.wikidata.org/wiki/Q111029","display_name":"Root (linguistics)","level":2,"score":0.2531999945640564}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.5281/zenodo.23978765","doi":"https://doi.org/10.5281/zenodo.23978765","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.23978765","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},{"id":"doi:10.5281/zenodo.23978766","doi":"https://doi.org/10.5281/zenodo.23978766","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.23978766","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.5281/zenodo.23978765","doi":"https://doi.org/10.5281/zenodo.23978765","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.23978765","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7225036455","https://openalex.org/W7204584025","https://openalex.org/W7223907446","https://openalex.org/W7212053715","https://openalex.org/W7225292269","https://openalex.org/W7170162272","https://openalex.org/W7165183485","https://openalex.org/W7204838952","https://openalex.org/W7167627487","https://openalex.org/W7110894099"],"abstract_inverted_index":{"Agents":[0],"that":[1,47,140],"read":[2],"files,":[3],"run":[4],"commands":[5],"and":[6,51,115,135,184,193],"send":[7],"mail":[8],"sometimes":[9],"do":[10],"more":[11,103],"than":[12,104],"asked,":[13],"with":[14,162,169],"no":[15],"attacker":[16],"involved.":[17],"We":[18],"measured":[19],"this":[20],"in":[21,33,39],"a":[22,48,52,58,67,88,122,146],"pre-registered":[23],"study:":[24],"three":[25],"Claude":[26],"5.5":[27],"models":[28],"ran":[29],"23":[30],"office":[31],"chores":[32],"an":[34,95],"executed":[35],"sandbox,":[36],"14,720":[37],"runs":[38,65,133],"four":[40],"frozen":[41],"stages.":[42],"The":[43,98],"main":[44],"finding":[45],"is":[46,60,125,141,187],"failing":[49,147],"tool":[50],"vague":[53,79,107],"prompt":[54],"go":[55],"together.":[56],"When":[57,121],"write":[59],"refused,":[61],"the":[62,113,126,137,149,177,194],"share":[63],"of":[64,90,145],"making":[66],"real":[68],"change":[69],"nobody":[70],"asked":[71],"for":[72],"rises":[73],"by":[74],"about":[75,82],"10":[76],"points":[77],"under":[78,85,189],"prompts":[80,108],"but":[81,168],"1":[83],"point":[84],"precise":[86],"ones,":[87],"comparison":[89],"two":[91],"registered":[92,178],"contrasts,":[93],"not":[94],"interaction":[96],"test.":[97],"agents":[99],"otherwise":[100],"look":[101],"far":[102],"they":[105],"act:":[106],"produce":[109],"much":[110],"reading":[111],"beyond":[112],"task,":[114],"unrequested":[116],"consequential":[117],"actions":[118],"are":[119],"rare.":[120],"privileged":[123],"command":[124],"only":[127,172],"way":[128],"to":[129],"finish,":[130],"almost":[131],"all":[132],"stop":[134],"tell":[136],"user;":[138],"whether":[139],"restraint":[142],"or":[143],"distrust":[144],"tool,":[148],"data":[150,192],"cannot":[151],"say.":[152],"Cheap":[153],"mitigations":[154],"barely":[155],"help.":[156],"Three":[157],"blind":[158],"model":[159],"raters":[160],"agreed":[161],"each":[163],"other":[164],"(Fleiss'":[165],"\u03ba":[166,174],"0.74)":[167],"our":[170],"rubric":[171],"at":[173],"0.59,":[175],"below":[176],"0.70,":[179],"so":[180],"it":[181],"was":[182],"revised":[183],"every":[185],"result":[186],"reported":[188],"both.":[190],"Code,":[191],"full":[195],"pre-registration":[196],"history:":[197],"https://github.com/m1ndl0ss/osar-bench":[198]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-11-06T00:00:00"}
