{"id":"557e1ce08a5a698fa5c691d3e40bc3611a2ce3cdc5220073721ed16b07d0799f","title":"STRIATUM-CTF uses typed tool calls to make autonomous CTF solving less brittle","argument":"The authors of STRIATUM-CTF argue that agents solving Capture-the-Flag (CTF) challenges need more than a language model sending commands to a shell: malformed commands, invented options and lost findings can derail long tasks. Their framework mediates between model reasoning and tool execution with the Model Context Protocol (MCP). Offensive-security tools are exposed as strictly typed, schema-validated functions; invalid calls are rejected before execution, and the agent receives structured observations it can use to self-correct.\n\nAcross 180 runs on 15 challenges, the system succeeded on 77.8–86.7% of attempts, with no significant difference between documentation conditions. In one live CTF, it placed first among more than 30 registered teams. The authors call that event a demonstration of feasibility, not superiority over human experts. Because every controlled-study condition used the same tool layer, the study tests documentation volume, not the protocol layer itself; the authors present schema-mediated grounding as a design principle with qualitative support rather than a measured effect.\n\nHow would a controlled comparison that varies the tool layer test whether schema-mediated grounding itself contributes to CTF-solving performance?","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The controlled study uses the same tool layer in all four conditions, so it does not measure the protocol layer's effect. The live CTF result comes from one non-replicated event and, as the authors note, does not establish superiority over human experts. The supplied evidence is the arXiv abstract and page text, not the full paper; details beyond that evidence are not assessed.","speaker":{"name":"Hugglestone, James, Chacko, Samuel Jacob, Stoller, Dawson, Schmidt, Ryan, Liu, Xiuwen","roleAtSource":"Article author"},"source":{"title":"STRIATUM-CTF: A Protocol-Driven Agentic Framework for Autonomous CTF Solving","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2603.22577","publishedAt":"2026-03-23","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors present STRIATUM-CTF as a framework that uses the Model Context Protocol (MCP) to mediate between LLM reasoning and tool execution.","excerpt":"We present STRIATUM-CTF, a modular agentic framework for Capture-the-Flag (CTF) solving that mediates between LLM reasoning and tool execution using the Model Context Protocol (MCP).","url":"https://arxiv.org/abs/2603.22577","relationship":"Source-stated"},{"text":"The framework exposes offensive-security tools as schema-validated functions, rejects calls that do not fit the schema, and returns structured observations for agent self-correction.","excerpt":"the protocol layer rejects out-of-schema tool calls before execution and returns structured observations that the agent uses to self-correct.","url":"https://arxiv.org/abs/2603.22577","relationship":"Source-stated"},{"text":"In a controlled study of 180 runs across 15 CTF challenges, success ranged from 77.8% to 86.7% across documentation conditions, with no significant differences between conditions.","excerpt":"In a controlled study of 180 runs across 15 CTF challenges under four levels of attached procedural documentation, the system succeeds on 77.8-86.7% of attempts; notably, a schema-only configuration with no procedural documentation already reaches 77.8%, and no condition differs significantly from a","url":"https://arxiv.org/abs/2603.22577","relationship":"Source-stated"},{"text":"The authors report that the system placed first at one live CTF, but characterize this single event as feasibility evidence rather than evidence of superiority over human experts.","excerpt":"In a single live, multi-team CTF at Florida State University, the system placed first against a field of more than 30 registered teams -- a single, non-replicated event decided by a 10-point margin over the runner-up, and therefore a demonstration of feasibility rather than of superiority over human","url":"https://arxiv.org/abs/2603.22577","relationship":"Source-stated"},{"text":"The authors say the controlled study isolates documentation volume, not the effect of the protocol layer, and describe schema-mediated grounding as a design principle with qualitative support.","excerpt":"Because all four ablation conditions share the same tool layer, our controlled study isolates documentation volume rather than the protocol layer itself; we accordingly present schema-mediated grounding as a design principle with qualitative support rather than as a measured effect.","url":"https://arxiv.org/abs/2603.22577","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2603.22577","checkedAt":"2026-11-11T12:06:07.685Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How would a controlled comparison that varies the tool layer test whether schema-mediated grounding itself contributes to CTF-solving performance?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"STRIATUM-CTF uses typed tool calls to make autonomous CTF solving less brittle","summary":"The authors of STRIATUM-CTF argue that agents solving Capture-the-Flag (CTF) challenges need more than a language model sending commands to a shell: malformed commands, invented options and lost findings can derail long tasks. Their framework mediates between model reasoning and tool execution with the Model Context Protocol (MCP). Offensive-security tools are exposed as strictly typed, schema-validated functions; invalid calls are rejected before execution, and the agent receives structured observations it can use to self-correct.\n\nAcross 180 runs on 15 challenges, the system succeeded on 77.8–86.7% of attempts, with no significant difference between documentation conditions. In one live CTF, it placed first among more than 30 registered teams. The authors call that event a demonstration of feasibility, not superiority over human experts. Because every controlled-study condition used the same tool layer, the study tests documentation volume, not the protocol layer itself; the authors present schema-mediated grounding as a design principle with qualitative support rather than a measured effect.\n\nHow would a controlled comparison that varies the tool layer test whether schema-mediated grounding itself contributes to CTF-solving performance?","question":"How would a controlled comparison that varies the tool layer test whether schema-mediated grounding itself contributes to CTF-solving performance?","claims":[{"text":"The authors present STRIATUM-CTF as a framework that uses the Model Context Protocol (MCP) to mediate between LLM reasoning and tool execution.","excerpt":"We present STRIATUM-CTF, a modular agentic framework for Capture-the-Flag (CTF) solving that mediates between LLM reasoning and tool execution using the Model Context Protocol (MCP).","support":"source-stated"},{"text":"The framework exposes offensive-security tools as schema-validated functions, rejects calls that do not fit the schema, and returns structured observations for agent self-correction.","excerpt":"the protocol layer rejects out-of-schema tool calls before execution and returns structured observations that the agent uses to self-correct.","support":"source-stated"},{"text":"In a controlled study of 180 runs across 15 CTF challenges, success ranged from 77.8% to 86.7% across documentation conditions, with no significant differences between conditions.","excerpt":"In a controlled study of 180 runs across 15 CTF challenges under four levels of attached procedural documentation, the system succeeds on 77.8-86.7% of attempts; notably, a schema-only configuration with no procedural documentation already reaches 77.8%, and no condition differs significantly from a","support":"source-stated"},{"text":"The authors report that the system placed first at one live CTF, but characterize this single event as feasibility evidence rather than evidence of superiority over human experts.","excerpt":"In a single live, multi-team CTF at Florida State University, the system placed first against a field of more than 30 registered teams -- a single, non-replicated event decided by a 10-point margin over the runner-up, and therefore a demonstration of feasibility rather than of superiority over human","support":"source-stated"},{"text":"The authors say the controlled study isolates documentation volume, not the effect of the protocol layer, and describe schema-mediated grounding as a design principle with qualitative support.","excerpt":"Because all four ablation conditions share the same tool layer, our controlled study isolates documentation volume rather than the protocol layer itself; we accordingly present schema-mediated grounding as a design principle with qualitative support rather than as a measured effect.","support":"source-stated"}],"limitations":["The controlled study uses the same tool layer in all four conditions, so it does not measure the protocol layer's effect.","The live CTF result comes from one non-replicated event and, as the authors note, does not establish superiority over human experts.","The supplied evidence is the arXiv abstract and page text, not the full paper; details beyond that evidence are not assessed."],"mediaId":"bafb93b9deedd30377346b75148aa882375c1a47c12a9a7ea15a49d80bd4b99f","correctionReason":null,"id":"557e1ce08a5a698fa5c691d3e40bc3611a2ce3cdc5220073721ed16b07d0799f","publicationId":"946bacbd-9f2a-4bea-85db-eeed07ddf99e","version":1,"sourceUrl":"https://arxiv.org/abs/2603.22577","sourceTitle":"STRIATUM-CTF: A Protocol-Driven Agentic Framework for Autonomous CTF Solving","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-03-23","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-02T05:04:05.804Z","createdAt":"2026-11-10T19:03:29.313Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"bafb93b9deedd30377346b75148aa882375c1a47c12a9a7ea15a49d80bd4b99f","kind":"article","url":"https://arxiv.org/abs/2603.22577","sourceUrl":"https://arxiv.org/abs/2603.22577","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-03T05:17:10.170Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2603.22577","checkedAt":"2026-11-11T12:06:07.685Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"bafb93b9deedd30377346b75148aa882375c1a47c12a9a7ea15a49d80bd4b99f","confidence":0.87,"provider":"typesafe","model":"jev-1.13.0","inputHash":"327ac7df171ce27d09d2b275c9204308530e819cb5987d1000d808fb8b9e0338","decidedAt":"2026-11-03T05:17:13.617Z"},"provenance":{"textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observationId":"a777d082b64a680dc864a60681bc3f0f8cc0476eaea123d427802d1c39f83148","retrievedAt":"2026-11-03T05:17:09.682Z","kind":"publisher-page","preparer":{"model":"server-excerpt-repair/stream-excerpt-repair-v1 (original draft: gpt-6-luna)","runId":"f6fa91cf-bc45-42dc-8cf8-091b44646169"},"reviewer":{"model":"gpt-6-luna","runId":"061ad8f6-68f6-4f6e-8346-d4f5ab1bb43d"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-10T19:03:29.313Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Hugglestone, James","url":null,"kind":"unknown"},{"name":"Chacko, Samuel Jacob","url":null,"kind":"unknown"},{"name":"Stoller, Dawson","url":null,"kind":"unknown"},{"name":"Schmidt, Ryan","url":null,"kind":"unknown"},{"name":"Liu, Xiuwen","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Hugglestone, James","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Hugglestone, James","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2603.22577","locator":"head/meta[name=author|citation_author]"},"assertionId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T12:06:07.685Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","sourceUrl":"https://arxiv.org/abs/2603.22577","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z"}},{"actor":{"name":"Chacko, Samuel Jacob","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Chacko, Samuel Jacob","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2603.22577","locator":"head/meta[name=author|citation_author]"},"assertionId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T12:06:07.685Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","sourceUrl":"https://arxiv.org/abs/2603.22577","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z"}},{"actor":{"name":"Stoller, Dawson","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Stoller, Dawson","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2603.22577","locator":"head/meta[name=author|citation_author]"},"assertionId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T12:06:07.685Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","sourceUrl":"https://arxiv.org/abs/2603.22577","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z"}},{"actor":{"name":"Schmidt, Ryan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Schmidt, Ryan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2603.22577","locator":"head/meta[name=author|citation_author]"},"assertionId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T12:06:07.685Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","sourceUrl":"https://arxiv.org/abs/2603.22577","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z"}},{"actor":{"name":"Liu, Xiuwen","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Liu, Xiuwen","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2603.22577","locator":"head/meta[name=author|citation_author]"},"assertionId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T12:06:07.685Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","sourceUrl":"https://arxiv.org/abs/2603.22577","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z"}}],"participationCoverage":{"total":5,"returned":5,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2603.22577","snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z","basis":"publisher-page-metadata","sourceLinks":47,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/557e1ce08a5a698fa5c691d3e40bc3611a2ce3cdc5220073721ed16b07d0799f/source"}},"canonicalUrl":"https://escapemeridian.com/readings/striatum-ctf-uses-typed-tool-calls-to-make-autonomous-ctf-557e1ce0","formats":{"html":"/readings/striatum-ctf-uses-typed-tool-calls-to-make-autonomous-ctf-557e1ce0","markdown":"/readings/striatum-ctf-uses-typed-tool-calls-to-make-autonomous-ctf-557e1ce0.md","json":"/readings/striatum-ctf-uses-typed-tool-calls-to-make-autonomous-ctf-557e1ce0.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"557e1ce08a5a698fa5c691d3e40bc3611a2ce3cdc5220073721ed16b07d0799f","revision":"946bacbd-9f2a-4bea-85db-eeed07ddf99e","revisionKind":"publication-id","title":"STRIATUM-CTF uses typed tool calls to make autonomous CTF solving less brittle","url":"https://escapemeridian.com/readings/striatum-ctf-uses-typed-tool-calls-to-make-autonomous-ctf-557e1ce0"},"source":{"type":"source","id":"https://arxiv.org/abs/2603.22577","revision":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","revisionKind":"sha256-source-text","title":"STRIATUM-CTF: A Protocol-Driven Agentic Framework for Autonomous CTF Solving","url":"https://arxiv.org/abs/2603.22577"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"557e1ce08a5a698fa5c691d3e40bc3611a2ce3cdc5220073721ed16b07d0799f","title":"STRIATUM-CTF uses typed tool calls to make autonomous CTF solving less brittle","argument":"The authors of STRIATUM-CTF argue that agents solving Capture-the-Flag (CTF) challenges need more than a language model sending commands to a shell: malformed commands, invented options and lost findings can derail long tasks. Their framework mediates between model reasoning and tool execution with the Model Context Protocol (MCP). Offensive-security tools are exposed as strictly typed, schema-validated functions; invalid calls are rejected before execution, and the agent receives structured observations it can use to self-correct.\n\nAcross 180 runs on 15 challenges, the system succeeded on 77.8–86.7% of attempts, with no significant difference between documentation conditions. In one live CTF, it placed first among more than 30 registered teams. The authors call that event a demonstration of feasibility, not superiority over human experts. Because every controlled-study condition used the same tool layer, the study tests documentation volume, not the protocol layer itself; the authors present schema-mediated grounding as a design principle with qualitative support rather than a measured effect.\n\nHow would a controlled comparison that varies the tool layer test whether schema-mediated grounding itself contributes to CTF-solving performance?","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The controlled study uses the same tool layer in all four conditions, so it does not measure the protocol layer's effect. The live CTF result comes from one non-replicated event and, as the authors note, does not establish superiority over human experts. The supplied evidence is the arXiv abstract and page text, not the full paper; details beyond that evidence are not assessed.","speaker":{"name":"Hugglestone, James, Chacko, Samuel Jacob, Stoller, Dawson, Schmidt, Ryan, Liu, Xiuwen","roleAtSource":"Article author"},"source":{"title":"STRIATUM-CTF: A Protocol-Driven Agentic Framework for Autonomous CTF Solving","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2603.22577","publishedAt":"2026-03-23","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors present STRIATUM-CTF as a framework that uses the Model Context Protocol (MCP) to mediate between LLM reasoning and tool execution.","excerpt":"We present STRIATUM-CTF, a modular agentic framework for Capture-the-Flag (CTF) solving that mediates between LLM reasoning and tool execution using the Model Context Protocol (MCP).","url":"https://arxiv.org/abs/2603.22577","relationship":"Source-stated"},{"text":"The framework exposes offensive-security tools as schema-validated functions, rejects calls that do not fit the schema, and returns structured observations for agent self-correction.","excerpt":"the protocol layer rejects out-of-schema tool calls before execution and returns structured observations that the agent uses to self-correct.","url":"https://arxiv.org/abs/2603.22577","relationship":"Source-stated"},{"text":"In a controlled study of 180 runs across 15 CTF challenges, success ranged from 77.8% to 86.7% across documentation conditions, with no significant differences between conditions.","excerpt":"In a controlled study of 180 runs across 15 CTF challenges under four levels of attached procedural documentation, the system succeeds on 77.8-86.7% of attempts; notably, a schema-only configuration with no procedural documentation already reaches 77.8%, and no condition differs significantly from a","url":"https://arxiv.org/abs/2603.22577","relationship":"Source-stated"},{"text":"The authors report that the system placed first at one live CTF, but characterize this single event as feasibility evidence rather than evidence of superiority over human experts.","excerpt":"In a single live, multi-team CTF at Florida State University, the system placed first against a field of more than 30 registered teams -- a single, non-replicated event decided by a 10-point margin over the runner-up, and therefore a demonstration of feasibility rather than of superiority over human","url":"https://arxiv.org/abs/2603.22577","relationship":"Source-stated"},{"text":"The authors say the controlled study isolates documentation volume, not the effect of the protocol layer, and describe schema-mediated grounding as a design principle with qualitative support.","excerpt":"Because all four ablation conditions share the same tool layer, our controlled study isolates documentation volume rather than the protocol layer itself; we accordingly present schema-mediated grounding as a design principle with qualitative support rather than as a measured effect.","url":"https://arxiv.org/abs/2603.22577","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2603.22577","checkedAt":"2026-11-11T12:06:07.685Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How would a controlled comparison that varies the tool layer test whether schema-mediated grounding itself contributes to CTF-solving performance?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"STRIATUM-CTF uses typed tool calls to make autonomous CTF solving less brittle","summary":"The authors of STRIATUM-CTF argue that agents solving Capture-the-Flag (CTF) challenges need more than a language model sending commands to a shell: malformed commands, invented options and lost findings can derail long tasks. Their framework mediates between model reasoning and tool execution with the Model Context Protocol (MCP). Offensive-security tools are exposed as strictly typed, schema-validated functions; invalid calls are rejected before execution, and the agent receives structured observations it can use to self-correct.\n\nAcross 180 runs on 15 challenges, the system succeeded on 77.8–86.7% of attempts, with no significant difference between documentation conditions. In one live CTF, it placed first among more than 30 registered teams. The authors call that event a demonstration of feasibility, not superiority over human experts. Because every controlled-study condition used the same tool layer, the study tests documentation volume, not the protocol layer itself; the authors present schema-mediated grounding as a design principle with qualitative support rather than a measured effect.\n\nHow would a controlled comparison that varies the tool layer test whether schema-mediated grounding itself contributes to CTF-solving performance?","question":"How would a controlled comparison that varies the tool layer test whether schema-mediated grounding itself contributes to CTF-solving performance?","claims":[{"text":"The authors present STRIATUM-CTF as a framework that uses the Model Context Protocol (MCP) to mediate between LLM reasoning and tool execution.","excerpt":"We present STRIATUM-CTF, a modular agentic framework for Capture-the-Flag (CTF) solving that mediates between LLM reasoning and tool execution using the Model Context Protocol (MCP).","support":"source-stated"},{"text":"The framework exposes offensive-security tools as schema-validated functions, rejects calls that do not fit the schema, and returns structured observations for agent self-correction.","excerpt":"the protocol layer rejects out-of-schema tool calls before execution and returns structured observations that the agent uses to self-correct.","support":"source-stated"},{"text":"In a controlled study of 180 runs across 15 CTF challenges, success ranged from 77.8% to 86.7% across documentation conditions, with no significant differences between conditions.","excerpt":"In a controlled study of 180 runs across 15 CTF challenges under four levels of attached procedural documentation, the system succeeds on 77.8-86.7% of attempts; notably, a schema-only configuration with no procedural documentation already reaches 77.8%, and no condition differs significantly from a","support":"source-stated"},{"text":"The authors report that the system placed first at one live CTF, but characterize this single event as feasibility evidence rather than evidence of superiority over human experts.","excerpt":"In a single live, multi-team CTF at Florida State University, the system placed first against a field of more than 30 registered teams -- a single, non-replicated event decided by a 10-point margin over the runner-up, and therefore a demonstration of feasibility rather than of superiority over human","support":"source-stated"},{"text":"The authors say the controlled study isolates documentation volume, not the effect of the protocol layer, and describe schema-mediated grounding as a design principle with qualitative support.","excerpt":"Because all four ablation conditions share the same tool layer, our controlled study isolates documentation volume rather than the protocol layer itself; we accordingly present schema-mediated grounding as a design principle with qualitative support rather than as a measured effect.","support":"source-stated"}],"limitations":["The controlled study uses the same tool layer in all four conditions, so it does not measure the protocol layer's effect.","The live CTF result comes from one non-replicated event and, as the authors note, does not establish superiority over human experts.","The supplied evidence is the arXiv abstract and page text, not the full paper; details beyond that evidence are not assessed."],"mediaId":"bafb93b9deedd30377346b75148aa882375c1a47c12a9a7ea15a49d80bd4b99f","correctionReason":null,"id":"557e1ce08a5a698fa5c691d3e40bc3611a2ce3cdc5220073721ed16b07d0799f","publicationId":"946bacbd-9f2a-4bea-85db-eeed07ddf99e","version":1,"sourceUrl":"https://arxiv.org/abs/2603.22577","sourceTitle":"STRIATUM-CTF: A Protocol-Driven Agentic Framework for Autonomous CTF Solving","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-03-23","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-02T05:04:05.804Z","createdAt":"2026-11-10T19:03:29.313Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"bafb93b9deedd30377346b75148aa882375c1a47c12a9a7ea15a49d80bd4b99f","kind":"article","url":"https://arxiv.org/abs/2603.22577","sourceUrl":"https://arxiv.org/abs/2603.22577","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-03T05:17:10.170Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2603.22577","checkedAt":"2026-11-11T12:06:07.685Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"bafb93b9deedd30377346b75148aa882375c1a47c12a9a7ea15a49d80bd4b99f","confidence":0.87,"provider":"typesafe","model":"jev-1.13.0","inputHash":"327ac7df171ce27d09d2b275c9204308530e819cb5987d1000d808fb8b9e0338","decidedAt":"2026-11-03T05:17:13.617Z"},"provenance":{"textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observationId":"a777d082b64a680dc864a60681bc3f0f8cc0476eaea123d427802d1c39f83148","retrievedAt":"2026-11-03T05:17:09.682Z","kind":"publisher-page","preparer":{"model":"server-excerpt-repair/stream-excerpt-repair-v1 (original draft: gpt-6-luna)","runId":"f6fa91cf-bc45-42dc-8cf8-091b44646169"},"reviewer":{"model":"gpt-6-luna","runId":"061ad8f6-68f6-4f6e-8346-d4f5ab1bb43d"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-10T19:03:29.313Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Hugglestone, James","url":null,"kind":"unknown"},{"name":"Chacko, Samuel Jacob","url":null,"kind":"unknown"},{"name":"Stoller, Dawson","url":null,"kind":"unknown"},{"name":"Schmidt, Ryan","url":null,"kind":"unknown"},{"name":"Liu, Xiuwen","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Hugglestone, James","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Hugglestone, James","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2603.22577","locator":"head/meta[name=author|citation_author]"},"assertionId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T12:06:07.685Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","sourceUrl":"https://arxiv.org/abs/2603.22577","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z"}},{"actor":{"name":"Chacko, Samuel Jacob","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Chacko, Samuel Jacob","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2603.22577","locator":"head/meta[name=author|citation_author]"},"assertionId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T12:06:07.685Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","sourceUrl":"https://arxiv.org/abs/2603.22577","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z"}},{"actor":{"name":"Stoller, Dawson","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Stoller, Dawson","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2603.22577","locator":"head/meta[name=author|citation_author]"},"assertionId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T12:06:07.685Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","sourceUrl":"https://arxiv.org/abs/2603.22577","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z"}},{"actor":{"name":"Schmidt, Ryan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Schmidt, Ryan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2603.22577","locator":"head/meta[name=author|citation_author]"},"assertionId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T12:06:07.685Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","sourceUrl":"https://arxiv.org/abs/2603.22577","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z"}},{"actor":{"name":"Liu, Xiuwen","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Liu, Xiuwen","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2603.22577","locator":"head/meta[name=author|citation_author]"},"assertionId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T12:06:07.685Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","sourceUrl":"https://arxiv.org/abs/2603.22577","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z"}}],"participationCoverage":{"total":5,"returned":5,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2603.22577","snapshotId":"2c830cbfc8032331205529ca13136e32200457f499c71e135070b0bc77cfcfda","contentSha256":"748dd61c2a422e50eefec9f826709f23fb9abe54a76b4add6aae4da5198cab57","textHash":"5c9933300285d98880518351668245293910fca794f059042ae7979df4bdbbfb","observedAt":"2026-11-11T12:06:07.685Z","basis":"publisher-page-metadata","sourceLinks":47,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/557e1ce08a5a698fa5c691d3e40bc3611a2ce3cdc5220073721ed16b07d0799f/source"}}}],"audits":[{"publicationId":"946bacbd-9f2a-4bea-85db-eeed07ddf99e","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The draft consistently attributes the framework, results and caveats to the paper's authors. It does not present their claims as established fact or as Escape Meridian's view.","excerpt":"Authors:James Hugglestone, Samuel Jacob Chacko, Dawson Stoller, Ryan Schmidt, Xiuwen Liu"},{"dimension":"summary","status":"uncertain","explanation":"The summary is faithful and keeps the authors' own hedges. But a non-specialist is never told that Capture-the-Flag is a cybersecurity hacking competition, or that the work concerns autonomous cyber operations. That framing is central to why the work matters. Smallest repair: add a plain-language gloss, e.g. CTF as competitive security challenges where teams break into deliberately vulnerable systems, and note the source's autonomous cyber operations framing. MCP and schema-validated should also get tooltips or a short gloss.","excerpt":"Autonomous cyber operations increasingly depend on agents that can chain reconnaissance, exploitation, and analysis across many tools while maintaining state over long task horizons."},{"dimension":"date","status":"pass","explanation":"originalPublished 2026-03-23 matches the v1 submission date. It is not the later v2 revision date or the November feed date.","excerpt":"[Submitted on 23 Mar 2026 (v1), last revised 29 Oct 2026 (this version, v2)]"},{"dimension":"support","status":"pass","explanation":"Each claim matches the abstract. The draft accurately reports the 77.8–86.7% range, the absence of significant differences, the first-place finish as a feasibility demonstration only, and the authors' admission that the protocol layer was not isolated. The question follows from that stated gap and asserts no unsupported benefit.","excerpt":"we accordingly present schema-mediated grounding as a design principle with qualitative support rather than as a measured effect."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper. It is the most direct access to the source.","excerpt":null}],"createdAt":"2026-11-10T19:03:29.313Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}