{"id":"https://openalex.org/W7204690097","doi":"https://doi.org/10.5281/zenodo.22173128","title":"Cognitive Harvard Architectures for AI Agent Perimeter Defense: Resolving the Confused Deputy Problem in Model Context Protocol via Capability-Based Access Control","display_name":"Cognitive Harvard Architectures for AI Agent Perimeter Defense: Resolving the Confused Deputy Problem in Model Context Protocol via Capability-Based Access Control","publication_year":2026,"publication_date":"2026-08-30","ids":{"openalex":"https://openalex.org/W7204690097","doi":"https://doi.org/10.5281/zenodo.22173128"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.22173128","doi":"https://doi.org/10.5281/zenodo.22173128","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.22173128","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.5281/zenodo.22173128","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5148621879","display_name":"Rudraneel Das","orcid":"https://orcid.org/0009-0009-6173-0262","observed_orcids":["https://orcid.org/0009-0009-6173-0262"]},"institutions":[{"id":"https://openalex.org/I170979836","display_name":"Jadavpur University","ror":"https://ror.org/02af4h012","country_code":"IN","type":"education","lineage":["https://openalex.org/I170979836"]}],"countries":["IN"],"is_corresponding":true,"raw_author_name":"Das, Rudraneel","raw_affiliation_strings":["Jadavpur University"],"raw_orcid":"https://orcid.org/0009-0009-6173-0262","affiliations":[{"raw_affiliation_string":"Jadavpur University","institution_ids":["https://openalex.org/I170979836"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5148621879"],"corresponding_institution_ids":["https://openalex.org/I170979836"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.8988999724388123,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.8988999724388123,"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/T11424","display_name":"Security and Verification in Computing","score":0.09759999811649323,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10927","display_name":"Access Control and Trust","score":0.0012000000569969416,"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/ai-agent-security","display_name":"AI agent security","score":0.8799999952316284},{"id":"https://openalex.org/keywords/indirect-prompt-injection","display_name":"indirect prompt injection","score":0.9039999842643738},{"id":"https://openalex.org/keywords/model-context-protocol","display_name":"Model Context Protocol","score":0.7889999747276306},{"id":"https://openalex.org/keywords/capability-based-access-control","display_name":"capability-based access control","score":0.9850000143051147},{"id":"https://openalex.org/keywords/autonomous-ai-agents","display_name":"autonomous AI agents","score":0.5400000214576721},{"id":"https://openalex.org/keywords/prompt-injection-detection","display_name":"prompt injection detection","score":0.531000018119812},{"id":"https://openalex.org/keywords/llm-security","display_name":"LLM security","score":0.3919999897480011}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7605999708175659},{"id":"https://openalex.org/C141141315","wikidata":"https://www.wikidata.org/wiki/Q2379942","display_name":"Guard (computer science)","level":2,"score":0.7035999894142151},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5099999904632568},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.49939998984336853},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.4456000030040741},{"id":"https://openalex.org/C169900460","wikidata":"https://www.wikidata.org/wiki/Q2200417","display_name":"Cognition","level":2,"score":0.3862999975681305},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.3774999976158142},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.37290000915527344},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.36410000920295715},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.35530000925064087},{"id":"https://openalex.org/C20854674","wikidata":"https://www.wikidata.org/wiki/Q4386060","display_name":"Cognitive architecture","level":3,"score":0.353300005197525},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.3418000042438507},{"id":"https://openalex.org/C161407221","wikidata":"https://www.wikidata.org/wiki/Q4382939","display_name":"Cognitive model","level":3,"score":0.31709998846054077},{"id":"https://openalex.org/C154556556","wikidata":"https://www.wikidata.org/wiki/Q192969","display_name":"Computer multitasking","level":2,"score":0.31200000643730164},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.3077000081539154},{"id":"https://openalex.org/C527821871","wikidata":"https://www.wikidata.org/wiki/Q228502","display_name":"Access control","level":2,"score":0.3019999861717224},{"id":"https://openalex.org/C35390924","wikidata":"https://www.wikidata.org/wiki/Q661075","display_name":"Metaprogramming","level":2,"score":0.27219998836517334},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.2711000144481659},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.26840001344680786},{"id":"https://openalex.org/C157170001","wikidata":"https://www.wikidata.org/wiki/Q4781507","display_name":"Applications of artificial intelligence","level":2,"score":0.25940001010894775},{"id":"https://openalex.org/C2776834041","wikidata":"https://www.wikidata.org/wiki/Q25346349","display_name":"Execution model","level":2,"score":0.2583000063896179},{"id":"https://openalex.org/C145804949","wikidata":"https://www.wikidata.org/wiki/Q478123","display_name":"Situation awareness","level":2,"score":0.2515000104904175},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.25029999017715454}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.5281/zenodo.22173128","doi":"https://doi.org/10.5281/zenodo.22173128","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.22173128","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.5281/zenodo.22173128","doi":"https://doi.org/10.5281/zenodo.22173128","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.22173128","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7204964789","https://openalex.org/W7204807916","https://openalex.org/W7128644961","https://openalex.org/W7165456472","https://openalex.org/W7154511665","https://openalex.org/W7203949808","https://openalex.org/W7206215733","https://openalex.org/W7134839784","https://openalex.org/W7202132333","https://openalex.org/W7163890102"],"abstract_inverted_index":{"Autonomous":[0],"artificial":[1],"intelligence":[2],"agents":[3],"executing":[4],"over":[5,134,208],"extensible":[6],"tool":[7],"interfaces":[8],"(such":[9],"as":[10],"Anthropic's":[11],"Model":[12],"Context":[13],"Protocol)":[14],"reproduce":[15],"the":[16,42,55,65,76,117],"classic":[17],"vulnerabilities":[18],"of":[19,120,148,151,202,238],"shared-memory":[20],"computer":[21,104],"architectures:":[22],"because":[23],"autoregressive":[24],"Transformers":[25],"ingest":[26],"program":[27],"instructions":[28],"and":[29,53,112,174,199,220,270],"untrusted":[30,142],"third-party":[31],"data":[32,86],"within":[33,259],"a":[34,68,100,260],"single":[35],"homogeneous":[36],"context":[37],"window,":[38],"they":[39],"suffer":[40],"from":[41,88,169],"Cognitive":[43,77],"Von":[44],"Neumann":[45],"Conflation.":[46],"Adversarial":[47],"observations":[48],"can":[49],"shift":[50],"attention":[51],"weights":[52],"hijack":[54],"agent's":[56],"execution":[57,91,146],"policy":[58],"via":[59,247],"Indirect":[60],"Prompt":[61,222],"Injection":[62],"(IPI),":[63],"turning":[64],"agent":[66,122,274],"into":[67],"Confused":[69],"Deputy.":[70],"In":[71],"this":[72],"treatise,":[73],"we":[74,115],"formulate":[75],"Harvard":[78],"Architecture,":[79],"an":[80,93,141,145,153,200,234],"architectural":[81],"paradigm":[82],"that":[83,140],"physically":[84],"decouples":[85],"ingestion":[87],"privileged":[89],"action":[90],"through":[92,99],"external,":[94],"capability-mediated":[95],"transport":[96],"perimeter.":[97],"Synthesized":[98],"multi-perspective":[101],"inquiry":[102],"spanning":[103],"systems":[105],"architecture,":[106],"AI":[107],"safety":[108],"red-teaming,":[109],"formal":[110],"methods,":[111],"enterprise":[113],"DevOps,":[114],"establish":[116],"theoretical":[118],"foundations":[119],"capability-based":[121],"security.":[123],"We":[124,157],"formally":[125],"prove":[126],"Theorem":[127],"1":[128],"(Deterministic":[129],"Non-Interference)":[130],"by":[131],"mathematical":[132],"induction":[133],"arbitrary":[135],"trajectory":[136],"length":[137],"T,":[138],"establishing":[139,204],"observation":[143],"has":[144],"probability":[147],"exactly":[149],"zero":[150],"triggering":[152],"out-of-scope":[154],"mutating":[155],"tool.":[156],"evaluate":[158],"Mastyf":[159,184,232],"Guard":[160,185,211,223],"across":[161],"50,000":[162],"balanced":[163],"empirical":[164],"samples":[165],"(25,000":[166],"adversarial":[167],"injections":[168],"UIUC":[170],"InjecAgent,":[171],"Microsoft":[172],"BIPIA,":[173],"NVIDIA":[175],"Garak":[176],"contrasted":[177],"against":[178],"25,000":[179],"authentic":[180],"developer":[181],"DevOps":[182],"commands).":[183],"achieves":[186,233],"99.33%":[187],"threat":[188],"recall":[189],"(95%":[190],"bootstrap":[191],"confidence":[192],"interval:":[193],"[99.22%,":[194],"99.43%],":[195],"B":[196],"=":[197,215,226],"1,000)":[198],"F1-score":[201],"0.9524,":[203],"statistically":[205],"conclusive":[206],"superiority":[207],"Meta":[209],"Llama":[210],"3":[212],"8B":[213],"(chi-square":[214,225],"7,156.1,":[216],"p":[217,228],"<":[218,229],"10^-15)":[219],"OpenAI":[221],"86M":[224],"11,496.9,":[227],"10^-15).":[230],"Crucially,":[231],"amortized":[235],"pipelined":[236],"latency":[237],"4.8":[239],"microseconds":[240],"(0.005":[241],"ms)":[242],"on":[243],"standard":[244],"CPU":[245],"hardware":[246],"deterministic":[248],"fast-path":[249],"short-circuiting,":[250],"while":[251],"standalone":[252],"neural":[253],"inference":[254],"executes":[255],"in":[256],"18.4":[257],"ms":[258],"1.1":[261],"GB":[262],"RAM":[263],"footprint,":[264],"eliminating":[265],"dedicated":[266],"GPU":[267],"infrastructure":[268],"requirements":[269],"saving":[271],"$6,000/year":[272],"per":[273],"instance.":[275]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-08-31T00:00:00"}