{"id":"https://openalex.org/W7204456034","doi":"https://doi.org/10.48550/arxiv.2608.25920","title":"Repair or Resample? Rethinking Failure Debugging in LLM Multi-Agent Systems","display_name":"Repair or Resample? Rethinking Failure Debugging in LLM Multi-Agent Systems","publication_year":2026,"publication_date":"2026-08-26","ids":{"openalex":"https://openalex.org/W7204456034","doi":"https://doi.org/10.48550/arxiv.2608.25920"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.25920","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.25920","pdf_url":"https://arxiv.org/pdf/2608.25920","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.25920","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5148469152","display_name":"Zhongwen Luan","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I66867065","display_name":"East China Normal University","ror":"https://ror.org/02n96ep67","country_code":"CN","type":"education","lineage":["https://openalex.org/I66867065"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Luan, Zhongwen","raw_affiliation_strings":["East China Normal University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"East China Normal University","institution_ids":["https://openalex.org/I66867065"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148447128","display_name":"Xiaoyu Zhang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Zhang, Xiaoyu","raw_affiliation_strings":["Nanyang Technological University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanyang Technological University","institution_ids":["https://openalex.org/I172675005"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148445087","display_name":"Ming Hu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I66867065","display_name":"East China Normal University","ror":"https://ror.org/02n96ep67","country_code":"CN","type":"education","lineage":["https://openalex.org/I66867065"]},{"id":"https://openalex.org/I79891267","display_name":"Singapore Management University","ror":"https://ror.org/050qmg959","country_code":"SG","type":"education","lineage":["https://openalex.org/I79891267"]}],"countries":["CN","SG"],"is_corresponding":false,"raw_author_name":"Hu, Ming","raw_affiliation_strings":["East China Normal University","Singapore Management University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"East China Normal University","institution_ids":["https://openalex.org/I66867065"]},{"raw_affiliation_string":"Singapore Management University","institution_ids":["https://openalex.org/I79891267"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148464880","display_name":"Yue Yang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I148099405","display_name":"Xi'an University of Architecture and Technology","ror":"https://ror.org/04v2j2k71","country_code":"CN","type":"education","lineage":["https://openalex.org/I148099405"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yang, Yue","raw_affiliation_strings":["Xi'an University of Architecture and Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xi'an University of Architecture and Technology","institution_ids":["https://openalex.org/I148099405"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5058013874","display_name":"Jiongchi Yu","orcid":"https://orcid.org/0000-0002-2888-4499","observed_orcids":["https://orcid.org/0000-0002-2888-4499"]},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]},{"id":"https://openalex.org/I79891267","display_name":"Singapore Management University","ror":"https://ror.org/050qmg959","country_code":"SG","type":"education","lineage":["https://openalex.org/I79891267"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Yu, Jiongchi","raw_affiliation_strings":["Nanyang Technological University","Singapore Management University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanyang Technological University","institution_ids":["https://openalex.org/I172675005"]},{"raw_affiliation_string":"Singapore Management University","institution_ids":["https://openalex.org/I79891267"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5148450324","display_name":"Xiaohong Chen","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I66867065","display_name":"East China Normal University","ror":"https://ror.org/02n96ep67","country_code":"CN","type":"education","lineage":["https://openalex.org/I66867065"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chen, Xiaohong","raw_affiliation_strings":["East China Normal University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"East China Normal University","institution_ids":["https://openalex.org/I66867065"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":6,"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/T10743","display_name":"Software Testing and Debugging Techniques","score":0.7339000105857849,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"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/T10743","display_name":"Software Testing and Debugging Techniques","score":0.7339000105857849,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"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/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.2012999951839447,"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/T12127","display_name":"Software System Performance and Reliability","score":0.019300000742077827,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/multi-agent-llm-systems","display_name":"multi-agent LLM systems","score":0.8389999866485596},{"id":"https://openalex.org/keywords/fault-correction","display_name":"fault repair","score":0.628000020980835},{"id":"https://openalex.org/keywords/failure-reproduction","display_name":"failure reproduction","score":0.7070000171661377},{"id":"https://openalex.org/keywords/long-horizon-tasks","display_name":"long-horizon tasks","score":0.6320000290870667}],"concepts":[{"id":"https://openalex.org/C168065819","wikidata":"https://www.wikidata.org/wiki/Q845566","display_name":"Debugging","level":2,"score":0.8690999746322632},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7350999712944031},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.48429998755455017},{"id":"https://openalex.org/C138673069","wikidata":"https://www.wikidata.org/wiki/Q322229","display_name":"Tracing","level":2,"score":0.4397999942302704},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.43869999051094055},{"id":"https://openalex.org/C75291252","wikidata":"https://www.wikidata.org/wiki/Q1315756","display_name":"TRACE (psycholinguistics)","level":2,"score":0.4140999913215637},{"id":"https://openalex.org/C150921843","wikidata":"https://www.wikidata.org/wiki/Q1170431","display_name":"Resampling","level":2,"score":0.3926999866962433},{"id":"https://openalex.org/C200601418","wikidata":"https://www.wikidata.org/wiki/Q2193887","display_name":"Reliability engineering","level":1,"score":0.36160001158714294},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.3555999994277954},{"id":"https://openalex.org/C125112378","wikidata":"https://www.wikidata.org/wiki/Q176640","display_name":"Randomness","level":2,"score":0.33480000495910645},{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.3262999951839447},{"id":"https://openalex.org/C180591934","wikidata":"https://www.wikidata.org/wiki/Q1253369","display_name":"Downtime","level":2,"score":0.31610000133514404},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.31470000743865967},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.30970001220703125},{"id":"https://openalex.org/C136388014","wikidata":"https://www.wikidata.org/wiki/Q17084976","display_name":"Algorithmic program debugging","level":3,"score":0.30799999833106995},{"id":"https://openalex.org/C2780615836","wikidata":"https://www.wikidata.org/wiki/Q2471869","display_name":"USable","level":2,"score":0.29019999504089355},{"id":"https://openalex.org/C26713055","wikidata":"https://www.wikidata.org/wiki/Q245962","display_name":"Implementation","level":2,"score":0.2847000062465668},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.2824000120162964},{"id":"https://openalex.org/C2778012447","wikidata":"https://www.wikidata.org/wiki/Q1034415","display_name":"Scope (computer science)","level":2,"score":0.28139999508857727},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.2800999879837036},{"id":"https://openalex.org/C91071405","wikidata":"https://www.wikidata.org/wiki/Q1413145","display_name":"Program slicing","level":3,"score":0.2791999876499176},{"id":"https://openalex.org/C152745839","wikidata":"https://www.wikidata.org/wiki/Q5438153","display_name":"Fault detection and isolation","level":3,"score":0.2766999900341034},{"id":"https://openalex.org/C174348530","wikidata":"https://www.wikidata.org/wiki/Q188635","display_name":"Bridging (networking)","level":2,"score":0.27649998664855957},{"id":"https://openalex.org/C2779662365","wikidata":"https://www.wikidata.org/wiki/Q5416694","display_name":"Event (particle physics)","level":2,"score":0.27549999952316284},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.2669000029563904},{"id":"https://openalex.org/C113843644","wikidata":"https://www.wikidata.org/wiki/Q901882","display_name":"Interface (matter)","level":4,"score":0.26420000195503235},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.2635999917984009},{"id":"https://openalex.org/C25343380","wikidata":"https://www.wikidata.org/wiki/Q277521","display_name":"Relation (database)","level":2,"score":0.26339998841285706},{"id":"https://openalex.org/C153876917","wikidata":"https://www.wikidata.org/wiki/Q899704","display_name":"Traceability","level":2,"score":0.2596000134944916},{"id":"https://openalex.org/C173801870","wikidata":"https://www.wikidata.org/wiki/Q201413","display_name":"Heuristic","level":2,"score":0.258899986743927}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.25920","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.25920","pdf_url":"https://arxiv.org/pdf/2608.25920","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.25920","doi":"https://doi.org/10.48550/arxiv.2608.25920","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.25920","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.25920","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.25920","pdf_url":"https://arxiv.org/pdf/2608.25920","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7204456034.pdf","grobid_xml":"https://content.openalex.org/works/W7204456034.grobid-xml"},"referenced_works_count":13,"referenced_works":["https://openalex.org/W1643342729","https://openalex.org/W3177813494","https://openalex.org/W4378509427","https://openalex.org/W4385965642","https://openalex.org/W4399836817","https://openalex.org/W4402733601","https://openalex.org/W4404389496","https://openalex.org/W4404396586","https://openalex.org/W6891850465","https://openalex.org/W6910824972","https://openalex.org/W7134819923","https://openalex.org/W7151843496","https://openalex.org/W7165385035"],"related_works":["https://openalex.org/W7222852239","https://openalex.org/W7166819471","https://openalex.org/W7131436133","https://openalex.org/W7172258627","https://openalex.org/W7161680277","https://openalex.org/W7111301101","https://openalex.org/W7160865650","https://openalex.org/W4416255023","https://openalex.org/W7157506569","https://openalex.org/W6910824972"],"abstract_inverted_index":{"As":[0],"large":[1],"language":[2],"model":[3],"(LLM)-based":[4],"multi-agent":[5,227],"systems":[6],"(MASs)":[7],"are":[8,162],"increasingly":[9],"applied":[10],"to":[11,48,202,209],"long-horizon":[12],"complex":[13],"tasks,":[14],"their":[15,24],"reliability":[16],"has":[17],"emerged":[18],"as":[19],"the":[20,39,64,71,86,99,102,110,115,124,184,196,220,223],"core":[21],"bottleneck":[22],"hindering":[23],"real-world":[25],"deployment.":[26],"Existing":[27],"MAS":[28,56,74,87,119,152,214],"debugging":[29,215],"and":[30,37,90,107,136,169,174,216],"repair":[31,55,61,75,170,204,217],"methods":[32,53,161],"typically":[33],"rely":[34],"on":[35,140],"rerunning":[36],"resampling":[38],"entire":[40],"execution":[41,88,100],"trajectory.":[42],"However,":[43],"a":[44,80,145,187],"fundamental":[45],"question":[46],"remains":[47],"be":[49],"answered:":[50],"do":[51],"these":[52,141,179],"causally":[54],"failures":[57],"or":[58],"merely":[59],"stochastically":[60],"by":[62],"leveraging":[63],"randomness":[65],"of":[66,73,118,186,195,226],"LLM":[67],"sampling?":[68],"To":[69],"evaluate":[70],"effectiveness":[72,185],"methods,":[76],"we":[77,143,181],"introduce":[78],"SymTrace,":[79],"controlled":[81],"evaluation":[82],"framework":[83],"that":[84,157],"records":[85],"trajectory":[89],"establishes":[91],"intervention":[92,189],"anchors.":[93],"During":[94],"replay,":[95],"it":[96],"effectively":[97],"reconstructs":[98],"before":[101],"anchor":[103],"using":[104],"recorded":[105],"logs":[106],"only":[108],"regenerates":[109],"downstream":[111],"trajectory,":[112],"thereby":[113],"enabling":[114],"reliable":[116],"reproduction":[117,168],"failures.":[120],"We":[121],"further":[122,182],"construct":[123],"dataset":[125],"SymFail,":[126],"comprising":[127],"536":[128],"human-annotated":[129],"failure":[130,167],"trajectories":[131],"with":[132],"graph-linked":[133],"locations,":[134],"categories,":[135],"trace":[137],"evidence.":[138],"Based":[139],"foundations,":[142],"conduct":[144],"large-scale":[146],"empirical":[147],"study":[148,207],"across":[149],"three":[150],"mainstream":[151],"frameworks.":[153],"Our":[154],"findings":[155],"reveal":[156],"existing":[158],"unguided":[159],"rerun":[160],"highly":[163],"unreliable,":[164],"exhibiting":[165],"low":[166],"rates":[171],"(only":[172],"67.97%":[173],"6.90%,":[175],"respectively).":[176],"Building":[177],"upon":[178],"findings,":[180],"explore":[183],"symptom-driven":[188],"method,":[190],"which":[191],"successfully":[192],"repairs":[193],"20.15%":[194],"failed":[197],"cases":[198],"(a":[199],"191.89%":[200],"improvement":[201],"state-of-the-art":[203],"methods).":[205],"This":[206],"aims":[208],"provide":[210],"actionable":[211],"insights":[212],"for":[213,222],"research,":[218],"paving":[219],"way":[221],"robust":[224],"deployment":[225],"systems.":[228]},"counts_by_year":[{"year":2026,"cited_by_count":6}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-08-28T00:00:00"}
