{"id":"https://openalex.org/W7080127675","doi":"https://doi.org/10.48550/arxiv.2509.03990","title":"Meta-Policy Reflexion: Reusable Reflective Memory and Rule Admissibility for Resource-Efficient LLM Agent","display_name":"Meta-Policy Reflexion: Reusable Reflective Memory and Rule Admissibility for Resource-Efficient LLM Agent","publication_year":2025,"publication_date":"2025-09-04","ids":{"openalex":"https://openalex.org/W7080127675","doi":"https://doi.org/10.48550/arxiv.2509.03990"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2509.03990","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.03990","pdf_url":"https://arxiv.org/pdf/2509.03990","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/2509.03990","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Wu, Chunlong","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I116953780","display_name":"Tongji University","ror":"https://ror.org/03rc6as71","country_code":"CN","type":"education","lineage":["https://openalex.org/I116953780"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wu, Chunlong","raw_affiliation_strings":["Tongji University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tongji University","institution_ids":["https://openalex.org/I116953780"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Luo, Ye","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I116953780","display_name":"Tongji University","ror":"https://ror.org/03rc6as71","country_code":"CN","type":"education","lineage":["https://openalex.org/I116953780"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Luo, Ye","raw_affiliation_strings":["Tongji University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tongji University","institution_ids":["https://openalex.org/I116953780"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Qu, Zhibo","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I116953780","display_name":"Tongji University","ror":"https://ror.org/03rc6as71","country_code":"CN","type":"education","lineage":["https://openalex.org/I116953780"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qu, Zhibo","raw_affiliation_strings":["Tongji University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tongji University","institution_ids":["https://openalex.org/I116953780"]}]},{"author_position":"last","author":{"id":null,"display_name":"Wang, Min","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I116953780","display_name":"Tongji University","ror":"https://ror.org/03rc6as71","country_code":"CN","type":"education","lineage":["https://openalex.org/I116953780"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wang, Min","raw_affiliation_strings":["Tongji University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tongji University","institution_ids":["https://openalex.org/I116953780"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I116953780"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":10,"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":true,"primary_topic":{"id":"https://openalex.org/T10906","display_name":"AI-based Problem Solving and Planning","score":0.5976999998092651,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.5976999998092651,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.3125,"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":0.0210999995470047,"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":0.8309999704360962},{"id":"https://openalex.org/keywords/reflective-memory","display_name":"reflective memory","score":0.7749999761581421},{"id":"https://openalex.org/keywords/cross-task-transfer","display_name":"cross-task transfer","score":0.6570000052452087},{"id":"https://openalex.org/keywords/resource-efficient-inference","display_name":"resource-efficient inference","score":0.4970000088214874},{"id":"https://openalex.org/keywords/alfworld","display_name":"ALFWorld","score":0.5299999713897705}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8130000233650208},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.6908000111579895},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.6826000213623047},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5934000015258789},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.5077999830245972},{"id":"https://openalex.org/C177606310","wikidata":"https://www.wikidata.org/wiki/Q5674297","display_name":"Adaptability","level":2,"score":0.492900013923645},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3982999920845032},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.3921000063419342},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.3880999982357025},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.3862000107765198},{"id":"https://openalex.org/C179518139","wikidata":"https://www.wikidata.org/wiki/Q5140297","display_name":"Coding (social sciences)","level":2,"score":0.3402000069618225},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.33889999985694885},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.32409998774528503},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3028999865055084},{"id":"https://openalex.org/C2780598303","wikidata":"https://www.wikidata.org/wiki/Q65921492","display_name":"Flexibility (engineering)","level":2,"score":0.29249998927116394},{"id":"https://openalex.org/C3746660","wikidata":"https://www.wikidata.org/wiki/Q1068763","display_name":"Rule of inference","level":2,"score":0.28839999437332153}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2509.03990","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.03990","pdf_url":"https://arxiv.org/pdf/2509.03990","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.2509.03990","doi":"https://doi.org/10.48550/arxiv.2509.03990","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2509.03990","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:2509.03990","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.03990","pdf_url":"https://arxiv.org/pdf/2509.03990","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/W7080127675.pdf","grobid_xml":"https://content.openalex.org/works/W7080127675.grobid-xml"},"referenced_works_count":5,"referenced_works":["https://openalex.org/W4406031343","https://openalex.org/W4407759090","https://openalex.org/W4412203333","https://openalex.org/W4415110204","https://openalex.org/W7166853358"],"related_works":["https://openalex.org/W4398794858","https://openalex.org/W6891945665","https://openalex.org/W7162660316","https://openalex.org/W7223351337","https://openalex.org/W7117762039","https://openalex.org/W7133222210","https://openalex.org/W7221817212","https://openalex.org/W7125931769","https://openalex.org/W4414360997","https://openalex.org/W7117332116"],"abstract_inverted_index":{"Large":[0],"language":[1],"model":[2,103],"(LLM)":[3],"agents":[4],"achieve":[5],"impressive":[6],"single-task":[7],"performance":[8],"but":[9,29,48],"commonly":[10],"exhibit":[11],"repeated":[12],"failures,":[13],"inefficient":[14],"exploration,":[15],"and":[16,53,77,91,116,133,135,167,188,191,197],"limited":[17],"cross-task":[18],"adaptability.":[19],"Existing":[20],"reflective":[21],"strategies":[22],"(e.g.,":[23],"Reflexion,":[24],"ReAct)":[25],"improve":[26],"per-episode":[27],"behavior":[28],"typically":[30],"produce":[31,45],"ephemeral,":[32],"task-specific":[33],"traces":[34],"that":[35,66,79,182],"are":[36],"not":[37],"reused":[38],"across":[39],"tasks.":[40],"Reinforcement-learning":[41],"based":[42],"alternatives":[43],"can":[44],"transferable":[46],"policies":[47],"require":[49],"substantial":[50],"parameter":[51],"updates":[52],"compute.":[54],"In":[55],"this":[56],"work":[57],"we":[58],"introduce":[59],"Meta-Policy":[60,74],"Reflexion":[61,172],"(MPR):":[62],"a":[63,71,140],"hybrid":[64],"framework":[65],"consolidates":[67],"LLM-generated":[68],"reflections":[69],"into":[70],"structured,":[72],"predicate-like":[73],"Memory":[75],"(MPM)":[76],"applies":[78],"memory":[80],"at":[81],"inference":[82],"time":[83],"through":[84],"two":[85],"complementary":[86],"mechanisms":[87,181],"soft":[88],"memory-guided":[89],"decoding":[90],"hard":[92],"rule":[93,174],"admissibility":[94,175],"checks(HAC).":[95],"MPR":[96],"(i)":[97],"externalizes":[98],"reusable":[99],"corrective":[100],"knowledge":[101],"without":[102],"weight":[104],"updates,":[105],"(ii)":[106],"enforces":[107],"domain":[108],"constraints":[109],"to":[110,171],"reduce":[111],"unsafe":[112],"or":[113],"invalid":[114],"actions,":[115],"(iii)":[117],"retains":[118],"the":[119,126,137,145,150,158],"adaptability":[120],"of":[121],"language-based":[122],"reflection.":[123],"We":[124,179],"formalize":[125],"MPM":[127],"representation,":[128],"present":[129],"algorithms":[130],"for":[131,195],"update":[132],"decoding,":[134],"validate":[136],"approach":[138],"in":[139,149,157,164],"text-based":[141],"agent":[142],"environment":[143],"following":[144],"experimental":[146],"protocol":[147],"described":[148],"provided":[151],"implementation":[152],"(AlfWorld-based).":[153],"Empirical":[154],"results":[155],"reported":[156],"supplied":[159],"material":[160],"indicate":[161],"consistent":[162],"gains":[163],"execution":[165],"accuracy":[166],"robustness":[168],"when":[169],"compared":[170],"baselines;":[173],"further":[176],"improves":[177],"stability.":[178],"analyze":[180],"explain":[183],"these":[184],"gains,":[185],"discuss":[186],"scalability":[187],"failure":[189],"modes,":[190],"outline":[192],"future":[193],"directions":[194],"multimodal":[196],"multi-agent":[198],"extensions.":[199]},"counts_by_year":[{"year":2026,"cited_by_count":9},{"year":2025,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2025-10-10T00:00:00"}