{"id":"https://openalex.org/W7223703133","doi":"https://doi.org/10.66977/xsci.2611.000r","title":"HarnessMM: Recursive Self-Improvement for Harness Token Efficiency","display_name":"HarnessMM: Recursive Self-Improvement for Harness Token Efficiency","publication_year":2026,"publication_date":"2026-11-06","ids":{"openalex":"https://openalex.org/W7223703133","doi":"https://doi.org/10.66977/xsci.2611.000r"},"language":"en","primary_location":{"id":"doi:10.66977/xsci.2611.000r","doi":"https://doi.org/10.66977/xsci.2611.000r","is_oa":true,"landing_page_url":"https://doi.org/10.66977/xsci.2611.000r","pdf_url":"https://xsci.com/preprint/doi/10.66977/xsci.2611.000r.pdf","source":null,"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"posted-content"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://xsci.com/preprint/doi/10.66977/xsci.2611.000r.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5159733674","display_name":"Yuhua Jiang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuhua Jiang","raw_affiliation_strings":["Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5159684323","display_name":"Chang Xu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1290206253","display_name":"Microsoft (United States)","ror":"https://ror.org/00d0nc645","country_code":"US","type":"company","lineage":["https://openalex.org/I1290206253"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chang Xu","raw_affiliation_strings":["Microsoft Research Asia (MSRA)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft Research Asia (MSRA)","institution_ids":["https://openalex.org/I1290206253"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135691544","display_name":"Feifei Gao","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Feifei Gao","raw_affiliation_strings":["Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5159697600","display_name":"Li Zhao","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1290206253","display_name":"Microsoft (United States)","ror":"https://ror.org/00d0nc645","country_code":"US","type":"company","lineage":["https://openalex.org/I1290206253"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Li Zhao","raw_affiliation_strings":["Microsoft Research Asia (MSRA)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft Research Asia (MSRA)","institution_ids":["https://openalex.org/I1290206253"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.44290000200271606,"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":0.44290000200271606,"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":0.2768999934196472,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.06530000269412994,"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/agent-harnesses","display_name":"agent harnesses","score":0.9559999704360962},{"id":"https://openalex.org/keywords/recursive-self-improvement","display_name":"recursive self-improvement","score":0.9470000267028809},{"id":"https://openalex.org/keywords/token-efficiency","display_name":"token efficiency","score":0.6890000104904175},{"id":"https://openalex.org/keywords/decision-graphs","display_name":"decision graphs","score":0.6389999985694885},{"id":"https://openalex.org/keywords/prompt-refinement","display_name":"prompt refinement","score":0.46399998664855957},{"id":"https://openalex.org/keywords/codex","display_name":"codices","score":0.49399998784065247},{"id":"https://openalex.org/keywords/deepseek","display_name":"DeepSeek","score":0.8650000095367432},{"id":"https://openalex.org/keywords/humaneval","display_name":"HumanEval","score":0.7760000228881836},{"id":"https://openalex.org/keywords/mbpp","display_name":"MBPP","score":0.8579999804496765}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7278000116348267},{"id":"https://openalex.org/C206588197","wikidata":"https://www.wikidata.org/wiki/Q846574","display_name":"Reuse","level":2,"score":0.6197999715805054},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.4787999987602234},{"id":"https://openalex.org/C204495577","wikidata":"https://www.wikidata.org/wiki/Q1205349","display_name":"Callback","level":2,"score":0.47769999504089355},{"id":"https://openalex.org/C160145156","wikidata":"https://www.wikidata.org/wiki/Q778586","display_name":"Executable","level":2,"score":0.47600001096725464},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.4551999866962433},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.3758000135421753},{"id":"https://openalex.org/C15569618","wikidata":"https://www.wikidata.org/wiki/Q3561421","display_name":"Liveness","level":2,"score":0.37119999527931213},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.33340001106262207},{"id":"https://openalex.org/C3309909","wikidata":"https://www.wikidata.org/wiki/Q864155","display_name":"Binary decision diagram","level":2,"score":0.325300008058548},{"id":"https://openalex.org/C2780882242","wikidata":"https://www.wikidata.org/wiki/Q14235582","display_name":"Literal (mathematical logic)","level":2,"score":0.3183000087738037},{"id":"https://openalex.org/C97399411","wikidata":"https://www.wikidata.org/wiki/Q825367","display_name":"Coin flipping","level":2,"score":0.30469998717308044},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.29789999127388},{"id":"https://openalex.org/C2781238097","wikidata":"https://www.wikidata.org/wiki/Q175026","display_name":"Object (grammar)","level":2,"score":0.2928999960422516},{"id":"https://openalex.org/C115988155","wikidata":"https://www.wikidata.org/wiki/Q3262192","display_name":"Decision problem","level":2,"score":0.29100000858306885},{"id":"https://openalex.org/C113843644","wikidata":"https://www.wikidata.org/wiki/Q901882","display_name":"Interface (matter)","level":4,"score":0.2904999852180481},{"id":"https://openalex.org/C177225278","wikidata":"https://www.wikidata.org/wiki/Q192674","display_name":"Factoring","level":2,"score":0.2833999991416931},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.2775999903678894},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2741999924182892},{"id":"https://openalex.org/C193702766","wikidata":"https://www.wikidata.org/wiki/Q1414548","display_name":"Concurrency","level":2,"score":0.2741999924182892},{"id":"https://openalex.org/C203479927","wikidata":"https://www.wikidata.org/wiki/Q5165939","display_name":"Controller (irrigation)","level":2,"score":0.27410000562667847},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.2680000066757202},{"id":"https://openalex.org/C46304622","wikidata":"https://www.wikidata.org/wiki/Q374814","display_name":"Certification","level":2,"score":0.25130000710487366}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.66977/xsci.2611.000r","doi":"https://doi.org/10.66977/xsci.2611.000r","is_oa":true,"landing_page_url":"https://doi.org/10.66977/xsci.2611.000r","pdf_url":"https://xsci.com/preprint/doi/10.66977/xsci.2611.000r.pdf","source":null,"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"posted-content"},{"id":"pmh:oai:arXiv.org:2610.32813","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.32813","pdf_url":"https://arxiv.org/pdf/2610.32813","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":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2610.32813","doi":"https://doi.org/10.48550/arxiv.2610.32813","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2610.32813","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":"doi:10.66977/xsci.2611.000r","doi":"https://doi.org/10.66977/xsci.2611.000r","is_oa":true,"landing_page_url":"https://doi.org/10.66977/xsci.2611.000r","pdf_url":"https://xsci.com/preprint/doi/10.66977/xsci.2611.000r.pdf","source":null,"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"posted-content"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7223703133.pdf","grobid_xml":"https://content.openalex.org/works/W7223703133.grobid-xml"},"referenced_works_count":34,"referenced_works":["https://openalex.org/W2158181145","https://openalex.org/W2798935874","https://openalex.org/W2988421999","https://openalex.org/W3195140297","https://openalex.org/W4304195432","https://openalex.org/W4308244910","https://openalex.org/W4378505261","https://openalex.org/W4378942217","https://openalex.org/W4387321091","https://openalex.org/W4387389783","https://openalex.org/W4387799916","https://openalex.org/W4389519226","https://openalex.org/W4402500863","https://openalex.org/W4402667024","https://openalex.org/W4402670850","https://openalex.org/W4402671607","https://openalex.org/W4402671835","https://openalex.org/W4403444045","https://openalex.org/W4404781496","https://openalex.org/W4404782660","https://openalex.org/W4412505633","https://openalex.org/W4412889810","https://openalex.org/W4415253970","https://openalex.org/W4415332588","https://openalex.org/W4415539539","https://openalex.org/W7123666280","https://openalex.org/W7129014600","https://openalex.org/W7131632368","https://openalex.org/W7162340582","https://openalex.org/W7163683658","https://openalex.org/W7204215770","https://openalex.org/W7212048886","https://openalex.org/W7213587037","https://openalex.org/W7223024034"],"related_works":["https://openalex.org/W7220698031","https://openalex.org/W7223024034","https://openalex.org/W7220699204","https://openalex.org/W7169776101","https://openalex.org/W7223111218","https://openalex.org/W7169541667","https://openalex.org/W7223552490","https://openalex.org/W7213587037","https://openalex.org/W7162109044","https://openalex.org/W7160954296"],"abstract_inverted_index":{"HarnessMinusMinus":[0],"(HarnessMM)":[1],"is":[2,138,229],"a":[3,89],"research":[4],"project":[5],"on":[6],"improving":[7],"the":[8,13,203],"efficiency":[9],"of":[10],"agent":[11,225],"harnesses,":[12],"software":[14],"coordinating":[15],"model":[16],"reasoning,":[17],"tool":[18,46,125],"execution,":[19],"and":[20,30,36,65,75,106,192,207,211],"feedback.":[21],"This":[22],"technical":[23],"report":[24,200],"consolidates":[25],"its":[26],"shared":[27],"framework,":[28],"implementations,":[29],"empirical":[31],"studies":[32,218],"across":[33,223],"coding,":[34],"tool-based,":[35],"visual-agent":[37,205],"settings.":[38,227],"Agent":[39],"harnesses":[40],"often":[41],"invoke":[42],"reasoning":[43,129,137],"after":[44],"each":[45],"result,":[47],"even":[48],"when":[49,135],"earlier":[50],"calls":[51],"have":[52],"already":[53],"determined":[54],"subsequent":[55],"actions.":[56],"HarnessMM":[57,147],"addresses":[58],"this":[59],"overhead":[60,210],"through":[61,123],"guarded":[62,220],"decision":[63,221],"reuse":[64,141,222],"recursive":[66],"self-improvement":[67],"(RSI).":[68],"RSI":[69],"refines":[70],"author":[71],"prompts,":[72],"context":[73],"settings,":[74],"adapter":[76],"rules":[77],"using":[78],"training":[79],"feedback,":[80,116],"with":[81],"validation":[82],"selecting":[83],"deployment":[84],"configurations.":[85],"At":[86],"inference":[87],"time,":[88],"controller":[90],"reuses":[91],"resolved":[92],"decisions":[93,113],"while":[94,117,153,173],"keeping":[95],"actor":[96,133],"weights":[97],"fixed.":[98],"Executable":[99],"Decision":[100,109],"Graphs":[101],"encode":[102],"actions,":[103],"arguments,":[104],"dependencies,":[105],"guards.":[107],"Guarded":[108],"Reuse":[110],"activates":[111],"eligible":[112],"from":[114,158,169,176],"observed":[115],"Certified":[118],"Callback":[119],"Substitution":[120],"executes":[121],"them":[122],"native":[124],"interfaces,":[126],"replacing":[127],"redundant":[128],"callbacks.":[130,198],"The":[131,199],"original":[132],"resumes":[134],"new":[136],"required":[139],"or":[140],"conditions":[142],"are":[143],"unmet.":[144],"On":[145],"HumanEval,":[146],"preserves":[148],"Codex":[149],"success":[150,167],"at":[151,231],"90.82\\%":[152],"reducing":[154],"tokens":[155,174],"per":[156],"task":[157],"66.72k":[159],"to":[160,171,178],"47.02k":[161],"(29.54\\%":[162],"saving).":[163,181],"For":[164],"DeepSeek":[165],"Harness,":[166],"improves":[168],"71.43\\%":[170],"85.71\\%":[172],"decrease":[175],"96.28k":[177],"11.62k":[179],"(87.93\\%":[180],"An":[182],"MBPP":[183],"ablation":[184],"shows":[185],"26.91\\%":[186],"lower":[187,194],"cost":[188,195],"than":[189,196],"single-use":[190],"graphs":[191],"19.52\\%":[193],"retained":[197],"also":[201],"summarizes":[202],"HarnessMM-V":[204],"study":[206],"documents":[208],"optimization":[209],"execution":[212],"time":[213],"where":[214],"measured.":[215],"Together,":[216],"these":[217],"investigate":[219],"diverse":[224],"harness":[226],"Code":[228],"available":[230],"https://github.com/YuhuaJiang2002/HarnessMM.":[232]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-11T12:16:26.678958","created_date":"2026-10-27T00:00:00"}