{"id":"https://openalex.org/W7204173965","doi":"https://doi.org/10.48550/arxiv.2608.22167","title":"MCP-Universe RL: A Framework for Training MCP Tool-Use Agents via Reinforcement Learning","display_name":"MCP-Universe RL: A Framework for Training MCP Tool-Use Agents via Reinforcement Learning","publication_year":2026,"publication_date":"2026-08-23","ids":{"openalex":"https://openalex.org/W7204173965","doi":"https://doi.org/10.48550/arxiv.2608.22167"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.22167","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.22167","pdf_url":"https://arxiv.org/pdf/2608.22167","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/2608.22167","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5114803621","display_name":"Ziyang Luo","orcid":"https://orcid.org/0000-0002-6037-0471","observed_orcids":["https://orcid.org/0000-0002-6037-0471","https://orcid.org/0009-0009-4522-6525"]},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Luo, Ziyang","raw_affiliation_strings":["SalesforceAIResearch/MCP-Universe"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SalesforceAIResearch/MCP-Universe","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148122183","display_name":"Yan Yang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yang, Yan","raw_affiliation_strings":["Salesforce AI Research mcp-universe "],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Salesforce AI Research mcp-universe ","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050075702","display_name":"Xiangru Jian","orcid":"https://orcid.org/0009-0004-7138-7078","observed_orcids":["https://orcid.org/0009-0004-7138-7078"]},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jian, Xiangru","raw_affiliation_strings":["SalesforceAIResearch/MCP-Universe"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SalesforceAIResearch/MCP-Universe","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5036764042","display_name":"Ziji Shi","orcid":"https://orcid.org/0000-0001-9398-6507","observed_orcids":["https://orcid.org/0000-0001-9398-6507"]},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Shi, Ziji","raw_affiliation_strings":["SalesforceAIResearch/MCP-Universe"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SalesforceAIResearch/MCP-Universe","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148089706","display_name":"Xiaoqiang Lin","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Lin, Xiaoqiang","raw_affiliation_strings":["SalesforceAIResearch/MCP-Universe"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SalesforceAIResearch/MCP-Universe","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5014937540","display_name":"Jun Hao Liew","orcid":"https://orcid.org/0000-0002-7538-6759","observed_orcids":["https://orcid.org/0000-0002-7538-6759"]},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Liew, Jun Hao","raw_affiliation_strings":["SalesforceAIResearch/MCP-Universe"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SalesforceAIResearch/MCP-Universe","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5042646536","display_name":"Silvio Savarese","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Savarese, Silvio","raw_affiliation_strings":["Salesforce AI Research mcp-universe "],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Salesforce AI Research mcp-universe ","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5125514534","display_name":"Junnan Li","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Li, Junnan","raw_affiliation_strings":["SalesforceAIResearch/MCP-Universe"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SalesforceAIResearch/MCP-Universe","institution_ids":["https://openalex.org/I4210155268"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I4210155268"],"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.923799991607666,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.923799991607666,"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/T13904","display_name":"Artificial Intelligence Applications","score":0.01269999984651804,"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":0.011800000444054604,"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/tool-use","display_name":"tool use","score":0.6710000038146973},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.9419999718666077},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"reinforcement learning","score":0.6940000057220459},{"id":"https://openalex.org/keywords/model-context-protocol","display_name":"Model Context Protocol","score":0.9739999771118164},{"id":"https://openalex.org/keywords/gpu-utilization","display_name":"GPU utilization","score":0.6990000009536743},{"id":"https://openalex.org/keywords/software-engineering-agents","display_name":"software engineering agents","score":0.625},{"id":"https://openalex.org/keywords/deep-research-agents","display_name":"deep research agents","score":0.7799999713897705}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.8023999929428101},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7795000076293945},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.6669999957084656},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6450999975204468},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.6154999732971191},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.5746999979019165},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5218999981880188},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.49079999327659607},{"id":"https://openalex.org/C2781018962","wikidata":"https://www.wikidata.org/wiki/Q5164884","display_name":"Container (type theory)","level":2,"score":0.43479999899864197},{"id":"https://openalex.org/C53833338","wikidata":"https://www.wikidata.org/wiki/Q1061424","display_name":"Context switch","level":2,"score":0.429500013589859},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.3926999866962433},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.37560001015663147},{"id":"https://openalex.org/C113843644","wikidata":"https://www.wikidata.org/wiki/Q901882","display_name":"Interface (matter)","level":4,"score":0.3626999855041504},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.35429999232292175},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.3521000146865845},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.32989999651908875},{"id":"https://openalex.org/C190793597","wikidata":"https://www.wikidata.org/wiki/Q189768","display_name":"Application layer","level":3,"score":0.321399986743927},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.3125999867916107},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.2736000120639801},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.27239999175071716},{"id":"https://openalex.org/C89505385","wikidata":"https://www.wikidata.org/wiki/Q47146","display_name":"User interface","level":2,"score":0.2574000060558319},{"id":"https://openalex.org/C206952183","wikidata":"https://www.wikidata.org/wiki/Q1193100","display_name":"Preemption","level":2,"score":0.2547999918460846},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.2522999942302704}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.22167","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.22167","pdf_url":"https://arxiv.org/pdf/2608.22167","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.2608.22167","doi":"https://doi.org/10.48550/arxiv.2608.22167","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.22167","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:2608.22167","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.22167","pdf_url":"https://arxiv.org/pdf/2608.22167","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/W7204173965.pdf","grobid_xml":"https://content.openalex.org/works/W7204173965.grobid-xml"},"referenced_works_count":1,"referenced_works":["https://openalex.org/W7128641923"],"related_works":["https://openalex.org/W7077885510","https://openalex.org/W7155090208","https://openalex.org/W4416064147","https://openalex.org/W7221762837","https://openalex.org/W7126215865","https://openalex.org/W4414804199","https://openalex.org/W7102398531","https://openalex.org/W7123362110","https://openalex.org/W7169775648","https://openalex.org/W7128786764"],"abstract_inverted_index":{"Reinforcement":[0],"learning":[1],"(RL)":[2],"has":[3],"become":[4],"an":[5,43,89,115,138,182],"effective":[6],"way":[7],"to":[8,56,106,163],"improve":[9,210],"the":[10,25,32,60,64,98,104,107,128,146,165,179,196],"tool-use":[11,205],"ability":[12],"of":[13,48,50,75],"large":[14],"language":[15],"models":[16],"(LLMs),":[17],"but":[18],"most":[19],"existing":[20,183],"RL":[21,86,184],"frameworks":[22],"stop":[23],"at":[24],"policy":[26],"update.":[27],"For":[28],"every":[29],"new":[30],"domain,":[31],"user":[33],"is":[34],"left":[35],"with":[36,121,186],"two":[37,129],"hard":[38],"systems":[39],"problems:":[40],"standing":[41],"up":[42],"isolated":[44],"environment":[45],"for":[46],"each":[47],"hundreds":[49],"concurrent":[51],"trajectories":[52,162],"and":[53,58,133,144,154,188,203,209],"connecting":[54],"it":[55],"training,":[57],"scheduling":[59],"rollout":[61],"so":[62,109],"that":[63,72,92,141],"GPU":[65,166],"stays":[66],"busy":[67,167],"across":[68,136],"long,":[69],"multi-turn":[70],"episodes":[71,169],"spend":[73],"much":[74],"their":[76],"time":[77],"stalled":[78],"on":[79,171,207],"slow":[80],"tool":[81,111],"calls.":[82],"We":[83],"present":[84],"MCP-Universe":[85],"(MCP-U":[87],"RL),":[88],"open-source":[90],"framework":[91],"takes":[93],"over":[94,149],"both.":[95],"It":[96,126],"uses":[97],"Model":[99],"Context":[100],"Protocol":[101],"(MCP)":[102],"as":[103,114],"interface":[105],"environment,":[108],"any":[110],"already":[112],"exposed":[113],"MCP":[116,147],"server":[117],"plugs":[118],"into":[119],"training":[120,175],"no":[122],"RL-specific":[123],"integration":[124],"code.":[125],"builds":[127],"missing":[130],"layers":[131],"once":[132],"reuses":[134],"them":[135],"domains:":[137],"environment-orchestration":[139],"layer":[140,157,176],"provisions,":[142],"isolates,":[143],"recycles":[145],"environments":[148],"a":[150,155],"pluggable":[151],"container":[152],"backend,":[153,185],"rollout-orchestration":[156],"whose":[158],"staged":[159],"pipeline":[160],"overlaps":[161],"keep":[164],"while":[168],"wait":[170],"tools.":[172],"A":[173],"backend-agnostic":[174],"then":[177],"applies":[178],"update":[180],"through":[181],"veRL":[187],"slime":[189],"integrations.":[190],"With":[191],"one":[192],"configuration,":[193],"changing":[194],"only":[195],"task":[197,211],"specification,":[198],"we":[199],"train":[200],"software-engineering,":[201],"deep-research,":[202],"general":[204],"agents":[206],"gpt-oss-20b":[208],"reward":[212],"in":[213],"all":[214],"three.":[215]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-08-26T00:00:00"}