{"id":"https://openalex.org/W7216595909","doi":null,"title":"Self-Evolving Harness on Multiple Tasks with the Agent as Its Own Optimizer","display_name":"Self-Evolving Harness on Multiple Tasks with the Agent as Its Own Optimizer","publication_year":2026,"publication_date":"2026-09-29","ids":{"openalex":"https://openalex.org/W7216595909"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.38372","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.38372","pdf_url":"https://arxiv.org/pdf/2609.38372","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"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.38372","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5121931459","display_name":"Qiankai Xu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Xu, Qiankai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5121931459"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":1,"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/T11975","display_name":"Evolutionary Algorithms and Applications","score":0.7348999977111816,"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/T11975","display_name":"Evolutionary Algorithms and Applications","score":0.7348999977111816,"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/T10260","display_name":"Software Engineering Research","score":0.16279999911785126,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.04650000110268593,"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/language-model-agents","display_name":"language model agents","score":0.7080000042915344},{"id":"https://openalex.org/keywords/recursive-self-improvement","display_name":"recursive self-improvement","score":0.902999997138977},{"id":"https://openalex.org/keywords/multi-task-learning","display_name":"multi-task learning","score":0.609000027179718},{"id":"https://openalex.org/keywords/continuous-training","display_name":"continuous training","score":0.6299999952316284},{"id":"https://openalex.org/keywords/out-of-distribution-generalization","display_name":"out-of-distribution generalization","score":0.8610000014305115},{"id":"https://openalex.org/keywords/prompt-engineering","display_name":"prompt engineering","score":0.5550000071525574},{"id":"https://openalex.org/keywords/tool-calling","display_name":"tool calling","score":0.574999988079071},{"id":"https://openalex.org/keywords/code-generation","display_name":"code generation","score":0.43700000643730164},{"id":"https://openalex.org/keywords/codex","display_name":"codices","score":0.33399999141693115},{"id":"https://openalex.org/keywords/independent-review","display_name":"independent review","score":0.8190000057220459}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7418000102043152},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.6220999956130981},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.5665000081062317},{"id":"https://openalex.org/C126042441","wikidata":"https://www.wikidata.org/wiki/Q1324888","display_name":"Frame (networking)","level":2,"score":0.529699981212616},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.5016999840736389},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.49480000138282776},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.45019999146461487},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.4311000108718872},{"id":"https://openalex.org/C2778770139","wikidata":"https://www.wikidata.org/wiki/Q1966904","display_name":"Solver","level":2,"score":0.41429999470710754},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.41200000047683716},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3878999948501587},{"id":"https://openalex.org/C2780428219","wikidata":"https://www.wikidata.org/wiki/Q16952335","display_name":"Cover (algebra)","level":2,"score":0.3215000033378601},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.3066999912261963},{"id":"https://openalex.org/C182306322","wikidata":"https://www.wikidata.org/wiki/Q1779371","display_name":"Order (exchange)","level":2,"score":0.30480000376701355},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.29809999465942383},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2953999936580658},{"id":"https://openalex.org/C3019612716","wikidata":"https://www.wikidata.org/wiki/Q730920","display_name":"Problem solver","level":2,"score":0.29019999504089355},{"id":"https://openalex.org/C198352243","wikidata":"https://www.wikidata.org/wiki/Q37105","display_name":"Line (geometry)","level":2,"score":0.287200003862381},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2766000032424927},{"id":"https://openalex.org/C199519371","wikidata":"https://www.wikidata.org/wiki/Q942695","display_name":"Source lines of code","level":3,"score":0.27570000290870667},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.27480000257492065},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.2653999924659729},{"id":"https://openalex.org/C774472","wikidata":"https://www.wikidata.org/wiki/Q6760393","display_name":"Margin (machine learning)","level":2,"score":0.2596000134944916},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.25949999690055847}],"mesh":[],"locations_count":1,"locations":[{"id":"pmh:oai:arXiv.org:2609.38372","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.38372","pdf_url":"https://arxiv.org/pdf/2609.38372","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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2609.38372","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.38372","pdf_url":"https://arxiv.org/pdf/2609.38372","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":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7216595909.pdf","grobid_xml":"https://content.openalex.org/works/W7216595909.grobid-xml"},"referenced_works_count":46,"referenced_works":["https://openalex.org/W2155277832","https://openalex.org/W2518108298","https://openalex.org/W2902456977","https://openalex.org/W2908510526","https://openalex.org/W2913340405","https://openalex.org/W2964121744","https://openalex.org/W3118781290","https://openalex.org/W4304195432","https://openalex.org/W4387389783","https://openalex.org/W4387561453","https://openalex.org/W4391555701","https://openalex.org/W4392828353","https://openalex.org/W4399655920","https://openalex.org/W4412505633","https://openalex.org/W4412889810","https://openalex.org/W4415252687","https://openalex.org/W4415257516","https://openalex.org/W4416176141","https://openalex.org/W4416411532","https://openalex.org/W6948264221","https://openalex.org/W7103754783","https://openalex.org/W7125134911","https://openalex.org/W7130718641","https://openalex.org/W7133596291","https://openalex.org/W7134847331","https://openalex.org/W7147383744","https://openalex.org/W7151843496","https://openalex.org/W7158100772","https://openalex.org/W7160900680","https://openalex.org/W7162340582","https://openalex.org/W7163027430","https://openalex.org/W7163160704","https://openalex.org/W7164008618","https://openalex.org/W7164848732","https://openalex.org/W7165016123","https://openalex.org/W7168361355","https://openalex.org/W7172337529","https://openalex.org/W7172378928","https://openalex.org/W7172533549","https://openalex.org/W7202165692","https://openalex.org/W7202196053","https://openalex.org/W7202216241","https://openalex.org/W7206192636","https://openalex.org/W7206206100","https://openalex.org/W7213359549","https://openalex.org/W7214002283"],"related_works":["https://openalex.org/W7163027430","https://openalex.org/W7202208313","https://openalex.org/W7172337529","https://openalex.org/W7211901482","https://openalex.org/W7220436573","https://openalex.org/W7167605568","https://openalex.org/W7202216241","https://openalex.org/W7206206100","https://openalex.org/W7221547609","https://openalex.org/W7204267154"],"abstract_inverted_index":{"A":[0],"harness":[1,53,61,82,129,189],"is":[2,62],"the":[3,55,75,78,98,103,107,113,118,121,128,168,188,192,195,199,206,214,220,226,229,238,243,261],"code":[4],"around":[5],"a":[6,34,46,51,59,81,85,92,184],"language-model":[7],"agent":[8],"that":[9,130,246,263],"organizes":[10],"prompts,":[11],"calls":[12],"tools,":[13],"manages":[14],"context,":[15],"and":[16,58,77,115,125,148,154,179,209,222,272],"controls":[17],"execution.":[18],"As":[19],"models":[20],"grow":[21],"stronger,":[22],"recent":[23],"work":[24,37],"has":[25],"begun":[26],"to":[27,95,250],"let":[28],"agents":[29],"improve":[30],"their":[31],"own":[32],"harnesses,":[33],"line":[35],"of":[36,80,88,106,194,237,260],"known":[38],"as":[39,112,117,171],"self-evolving":[40],"harnesses.":[41],"In":[42,228],"most":[43],"existing":[44],"methods,":[45],"separate":[47,60],"proposer":[48],"running":[49],"on":[50,102,158,205,213,219,225,234,245],"human-designed":[52],"modifies":[54],"solver's":[56],"harness,":[57,108,187],"evolved":[63],"for":[64],"each":[65],"benchmark.":[66],"Real-world":[67],"tasks":[68,111,137,150],"come":[69],"from":[70,138,183,248],"many":[71],"domains,":[72],"so":[73],"both":[74],"evolution":[76,134,169,233],"evaluation":[79],"should":[83],"cover":[84],"diverse":[86],"range":[87],"tasks.":[89],"We":[90,166,254],"propose":[91],"framework":[93],"close":[94],"recursive":[96],"self-improvement:":[97],"same":[99,104],"frozen":[100],"model,":[101],"version":[105],"first":[109,196],"solves":[110],"solver":[114],"then,":[116],"proposer,":[119],"reads":[120],"complete":[122],"run":[123],"records":[124],"directly":[126],"edits":[127],"runs":[131],"it.":[132],"Each":[133],"batch":[135],"draws":[136],"five":[139,159],"benchmarks":[140,161,208],"in":[141],"different":[142],"domains.":[143],"To":[144],"measure":[145],"generalization,":[146],"training":[147,173],"held-out":[149],"are":[151],"strictly":[152],"separated,":[153],"we":[155],"additionally":[156],"evaluate":[157],"out-of-distribution":[160,215,239],"never":[162],"used":[163],"during":[164,265],"evolution.":[165],"frame":[167],"process":[170],"deep-learning":[172],"with":[174],"two":[175],"stages,":[176],"multi-task":[177],"pretraining":[178],"continual":[180],"training.":[181],"Starting":[182],"49-line":[185],"seed":[186],"obtained":[190],"at":[191],"end":[193],"stage":[197],"improves":[198],"average":[200],"score":[201,244],"by":[202,210],"4.48":[203],"points":[204,212],"in-distribution":[207],"12.64":[211],"benchmarks,":[216,240],"surpassing":[217],"Codex":[218],"former":[221],"matching":[223],"it":[224],"latter.":[227],"second":[230],"stage,":[231],"continued":[232],"Claw-Eval,":[235],"one":[236],"further":[241],"raises":[242],"benchmark":[247],"66.17":[249],"68.06,":[251],"exceeding":[252],"Codex.":[253],"also":[255],"provide":[256],"an":[257],"in-depth":[258],"analysis":[259],"mechanisms":[262],"emerged":[264],"evolution,":[266],"including":[267],"output":[268],"truncation,":[269],"history":[270],"compaction,":[271],"independent":[273],"review.":[274]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-10-03T00:00:00"}