{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:QUVXACU53YBLXZRF7HYQJWKV6S","short_pith_number":"pith:QUVXACU5","schema_version":"1.0","canonical_sha256":"852b700a9dde02bbe625f9f104d955f4b94a43a7148bd7f26c3c0fd70364170c","source":{"kind":"arxiv","id":"2510.01359","version":2},"attestation_state":"computed","paper":{"title":"Breaking the Code: Security Assessment of AI Code Agents Through Systematic Jailbreaking Attacks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Jifan Chen, Sam Mayers, Sanjay Krishna Gouda, Shoumik Saha, Varun Kumar, Zijian Wang","submitted_at":"2025-10-01T18:38:20Z","abstract_excerpt":"Code-capable large language model (LLM) agents are embedded in software engineering workflows where they can read, write, and execute code, raising \"jailbreak\" stakes beyond text-only settings. Prior evaluations emphasize refusal or harmful-text detection, leaving open whether agents compile and run malicious programs. We present JAWS-Bench (Jailbreaks Across WorkSpaces), a benchmark spanning three escalating workspace regimes mirroring attacker capability: empty (JAWS-0), single-file (JAWS-1), and multi-file (JAWS-M). We pair this with a hierarchical, executable-aware Judge Framework that tes"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2510.01359","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2025-10-01T18:38:20Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"69ab896ea1e4bf9c90702ddfe1a6637754d77e5f2725b81e1347548d87f0060d","abstract_canon_sha256":"a38aa80ad942eee57e41519637ac334eb55d1fd181d4a78fbd93c57c6b0e82fb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:09:51.423142Z","signature_b64":"/LapSRuv99cOTMyFe53wxNRqSN14Bqyq/oTqpLwzqlKrJiocjf0fdldcT/SCQ+sqyRzneapdvczIDZ/vOVy8Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"852b700a9dde02bbe625f9f104d955f4b94a43a7148bd7f26c3c0fd70364170c","last_reissued_at":"2026-06-19T16:09:51.422744Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:09:51.422744Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Breaking the Code: Security Assessment of AI Code Agents Through Systematic Jailbreaking Attacks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Jifan Chen, Sam Mayers, Sanjay Krishna Gouda, Shoumik Saha, Varun Kumar, Zijian Wang","submitted_at":"2025-10-01T18:38:20Z","abstract_excerpt":"Code-capable large language model (LLM) agents are embedded in software engineering workflows where they can read, write, and execute code, raising \"jailbreak\" stakes beyond text-only settings. Prior evaluations emphasize refusal or harmful-text detection, leaving open whether agents compile and run malicious programs. We present JAWS-Bench (Jailbreaks Across WorkSpaces), a benchmark spanning three escalating workspace regimes mirroring attacker capability: empty (JAWS-0), single-file (JAWS-1), and multi-file (JAWS-M). We pair this with a hierarchical, executable-aware Judge Framework that tes"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.01359","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2510.01359/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2510.01359","created_at":"2026-06-19T16:09:51.422804+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.01359v2","created_at":"2026-06-19T16:09:51.422804+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.01359","created_at":"2026-06-19T16:09:51.422804+00:00"},{"alias_kind":"pith_short_12","alias_value":"QUVXACU53YBL","created_at":"2026-06-19T16:09:51.422804+00:00"},{"alias_kind":"pith_short_16","alias_value":"QUVXACU53YBLXZRF","created_at":"2026-06-19T16:09:51.422804+00:00"},{"alias_kind":"pith_short_8","alias_value":"QUVXACU5","created_at":"2026-06-19T16:09:51.422804+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2605.20351","citing_title":"Refusal Evaluation in Coding LLMs and Code Agents: A Systematic Review of Thirteen Malicious-Code Prompt Corpora (2023-2025)","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14859","citing_title":"Do Coding Agents Understand Least-Privilege Authorization?","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10779","citing_title":"LITMUS: Benchmarking Behavioral Jailbreaks of LLM Agents in Real OS Environments","ref_index":37,"is_internal_anchor":true},{"citing_arxiv_id":"2605.03952","citing_title":"MOSAIC-Bench: Measuring Compositional Vulnerability Induction in Coding Agents","ref_index":9,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QUVXACU53YBLXZRF7HYQJWKV6S","json":"https://pith.science/pith/QUVXACU53YBLXZRF7HYQJWKV6S.json","graph_json":"https://pith.science/api/pith-number/QUVXACU53YBLXZRF7HYQJWKV6S/graph.json","events_json":"https://pith.science/api/pith-number/QUVXACU53YBLXZRF7HYQJWKV6S/events.json","paper":"https://pith.science/paper/QUVXACU5"},"agent_actions":{"view_html":"https://pith.science/pith/QUVXACU53YBLXZRF7HYQJWKV6S","download_json":"https://pith.science/pith/QUVXACU53YBLXZRF7HYQJWKV6S.json","view_paper":"https://pith.science/paper/QUVXACU5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.01359&json=true","fetch_graph":"https://pith.science/api/pith-number/QUVXACU53YBLXZRF7HYQJWKV6S/graph.json","fetch_events":"https://pith.science/api/pith-number/QUVXACU53YBLXZRF7HYQJWKV6S/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QUVXACU53YBLXZRF7HYQJWKV6S/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QUVXACU53YBLXZRF7HYQJWKV6S/action/storage_attestation","attest_author":"https://pith.science/pith/QUVXACU53YBLXZRF7HYQJWKV6S/action/author_attestation","sign_citation":"https://pith.science/pith/QUVXACU53YBLXZRF7HYQJWKV6S/action/citation_signature","submit_replication":"https://pith.science/pith/QUVXACU53YBLXZRF7HYQJWKV6S/action/replication_record"}},"created_at":"2026-06-19T16:09:51.422804+00:00","updated_at":"2026-06-19T16:09:51.422804+00:00"}