{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:FKCI5AICXHF4QQQU6M4SKZ6R6S","short_pith_number":"pith:FKCI5AIC","schema_version":"1.0","canonical_sha256":"2a848e8102b9cbc84214f3392567d1f49dc0160507334fa526ece93267e35048","source":{"kind":"arxiv","id":"2508.03936","version":1},"attestation_state":"computed","paper":{"title":"ASTRA: Autonomous Spatial-Temporal Red-teaming for AI Software Assistants","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG","cs.SE"],"primary_cat":"cs.CR","authors_text":"Chengpeng Wang, Guangyu Shen, Hanxi Guo, Jiasheng Jiang, Lu Yan, Siyuan Cheng, Xiangyu Zhang, Xiangzhe Xu, Xiaolong Jin, Xuan Chen, Zhuo Zhang, Zian Su","submitted_at":"2025-08-05T21:57:52Z","abstract_excerpt":"AI coding assistants like GitHub Copilot are rapidly transforming software development, but their safety remains deeply uncertain-especially in high-stakes domains like cybersecurity. Current red-teaming tools often rely on fixed benchmarks or unrealistic prompts, missing many real-world vulnerabilities. We present ASTRA, an automated agent system designed to systematically uncover safety flaws in AI-driven code generation and security guidance systems. ASTRA works in three stages: (1) it builds structured domain-specific knowledge graphs that model complex software tasks and known weaknesses;"},"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":"2508.03936","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-08-05T21:57:52Z","cross_cats_sorted":["cs.CL","cs.LG","cs.SE"],"title_canon_sha256":"3207532dc57e05934a32f75a4edeaf9e7dcb4f553c672697891ca0aee2a1bbcf","abstract_canon_sha256":"5d02bbb54bda5077b375fe3226ab1816c1d41810f688692a2ef00eac2a904a61"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:49:17.656225Z","signature_b64":"6pdWKHot6p+KSwuOo5WYpRguMq3rh+Xbm3uTQS7CvHr4PyycTa7W5JehDbUPSdRnfRLdr0lYngf/4jH4si4QCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2a848e8102b9cbc84214f3392567d1f49dc0160507334fa526ece93267e35048","last_reissued_at":"2026-07-05T11:49:17.655736Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:49:17.655736Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ASTRA: Autonomous Spatial-Temporal Red-teaming for AI Software Assistants","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG","cs.SE"],"primary_cat":"cs.CR","authors_text":"Chengpeng Wang, Guangyu Shen, Hanxi Guo, Jiasheng Jiang, Lu Yan, Siyuan Cheng, Xiangyu Zhang, Xiangzhe Xu, Xiaolong Jin, Xuan Chen, Zhuo Zhang, Zian Su","submitted_at":"2025-08-05T21:57:52Z","abstract_excerpt":"AI coding assistants like GitHub Copilot are rapidly transforming software development, but their safety remains deeply uncertain-especially in high-stakes domains like cybersecurity. Current red-teaming tools often rely on fixed benchmarks or unrealistic prompts, missing many real-world vulnerabilities. We present ASTRA, an automated agent system designed to systematically uncover safety flaws in AI-driven code generation and security guidance systems. ASTRA works in three stages: (1) it builds structured domain-specific knowledge graphs that model complex software tasks and known weaknesses;"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.03936","kind":"arxiv","version":1},"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/2508.03936/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":"2508.03936","created_at":"2026-07-05T11:49:17.655797+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.03936v1","created_at":"2026-07-05T11:49:17.655797+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.03936","created_at":"2026-07-05T11:49:17.655797+00:00"},{"alias_kind":"pith_short_12","alias_value":"FKCI5AICXHF4","created_at":"2026-07-05T11:49:17.655797+00:00"},{"alias_kind":"pith_short_16","alias_value":"FKCI5AICXHF4QQQU","created_at":"2026-07-05T11:49:17.655797+00:00"},{"alias_kind":"pith_short_8","alias_value":"FKCI5AIC","created_at":"2026-07-05T11:49:17.655797+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.28734","citing_title":"Code as a Weapon: A Consensus-Labeled Prompt Bank for Measuring Coding-Model Compliance with Malicious-Code Requests","ref_index":7,"is_internal_anchor":false},{"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":14,"is_internal_anchor":false},{"citing_arxiv_id":"2512.00380","citing_title":"Knowledge-Graph-Driven Data Synthesis for Low-Resource Software Development: A HarmonyOS Case Study","ref_index":70,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FKCI5AICXHF4QQQU6M4SKZ6R6S","json":"https://pith.science/pith/FKCI5AICXHF4QQQU6M4SKZ6R6S.json","graph_json":"https://pith.science/api/pith-number/FKCI5AICXHF4QQQU6M4SKZ6R6S/graph.json","events_json":"https://pith.science/api/pith-number/FKCI5AICXHF4QQQU6M4SKZ6R6S/events.json","paper":"https://pith.science/paper/FKCI5AIC"},"agent_actions":{"view_html":"https://pith.science/pith/FKCI5AICXHF4QQQU6M4SKZ6R6S","download_json":"https://pith.science/pith/FKCI5AICXHF4QQQU6M4SKZ6R6S.json","view_paper":"https://pith.science/paper/FKCI5AIC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.03936&json=true","fetch_graph":"https://pith.science/api/pith-number/FKCI5AICXHF4QQQU6M4SKZ6R6S/graph.json","fetch_events":"https://pith.science/api/pith-number/FKCI5AICXHF4QQQU6M4SKZ6R6S/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FKCI5AICXHF4QQQU6M4SKZ6R6S/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FKCI5AICXHF4QQQU6M4SKZ6R6S/action/storage_attestation","attest_author":"https://pith.science/pith/FKCI5AICXHF4QQQU6M4SKZ6R6S/action/author_attestation","sign_citation":"https://pith.science/pith/FKCI5AICXHF4QQQU6M4SKZ6R6S/action/citation_signature","submit_replication":"https://pith.science/pith/FKCI5AICXHF4QQQU6M4SKZ6R6S/action/replication_record"}},"created_at":"2026-07-05T11:49:17.655797+00:00","updated_at":"2026-07-05T11:49:17.655797+00:00"}