{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KCG2TGZ7QLOP7SH3SZE7X24JRC","short_pith_number":"pith:KCG2TGZ7","schema_version":"1.0","canonical_sha256":"508da99b3f82dcffc8fb9649fbeb8988ab474342da33bae742e40896b219c08f","source":{"kind":"arxiv","id":"2605.24171","version":1},"attestation_state":"computed","paper":{"title":"PromptAudit: Auditing Prompt Sensitivity in LLM-Based Vulnerability Detection","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"David Mohaisen, Mandana Ghadamian, Steffen J. Camarato, Yahya Hmaiti","submitted_at":"2026-05-22T19:44:51Z","abstract_excerpt":"Large language models are increasingly used for vulnerability detection, yet their reliability under different prompt formulations remains uncharacterized. We present PromptAudit, a controlled evaluation framework that isolates prompt effects by fixing the dataset, decoding, and parsing while varying only the prompting strategy. Using five prompting strategies across five open-weight models on 1,000 CVEs (6,074 code samples spanning 16 programming languages), we evaluate accuracy, recall, abstention, coverage, and effective F1. We find that standard chain-of-thought prompting achieves the stro"},"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":"2605.24171","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-22T19:44:51Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2262851f06c907e17a20bf68f74fc8560099add208056ab1f386f223fa5e24bb","abstract_canon_sha256":"4e39fbd2e75c15d11a8f00a56bf8226c8004a4b865a5377ef653ae259cf9b206"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:02:50.439691Z","signature_b64":"0wqHJTCr+pPYAJSmykcLs364T+XZIELSOCUp0NOQod+IG23upT0KBpUMpAD6zXhIROTfA0tgsDhcSia7QDi6BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"508da99b3f82dcffc8fb9649fbeb8988ab474342da33bae742e40896b219c08f","last_reissued_at":"2026-05-26T01:02:50.438863Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:02:50.438863Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PromptAudit: Auditing Prompt Sensitivity in LLM-Based Vulnerability Detection","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"David Mohaisen, Mandana Ghadamian, Steffen J. Camarato, Yahya Hmaiti","submitted_at":"2026-05-22T19:44:51Z","abstract_excerpt":"Large language models are increasingly used for vulnerability detection, yet their reliability under different prompt formulations remains uncharacterized. We present PromptAudit, a controlled evaluation framework that isolates prompt effects by fixing the dataset, decoding, and parsing while varying only the prompting strategy. Using five prompting strategies across five open-weight models on 1,000 CVEs (6,074 code samples spanning 16 programming languages), we evaluate accuracy, recall, abstention, coverage, and effective F1. We find that standard chain-of-thought prompting achieves the stro"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.24171","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/2605.24171/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":"2605.24171","created_at":"2026-05-26T01:02:50.439021+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.24171v1","created_at":"2026-05-26T01:02:50.439021+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.24171","created_at":"2026-05-26T01:02:50.439021+00:00"},{"alias_kind":"pith_short_12","alias_value":"KCG2TGZ7QLOP","created_at":"2026-05-26T01:02:50.439021+00:00"},{"alias_kind":"pith_short_16","alias_value":"KCG2TGZ7QLOP7SH3","created_at":"2026-05-26T01:02:50.439021+00:00"},{"alias_kind":"pith_short_8","alias_value":"KCG2TGZ7","created_at":"2026-05-26T01:02:50.439021+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KCG2TGZ7QLOP7SH3SZE7X24JRC","json":"https://pith.science/pith/KCG2TGZ7QLOP7SH3SZE7X24JRC.json","graph_json":"https://pith.science/api/pith-number/KCG2TGZ7QLOP7SH3SZE7X24JRC/graph.json","events_json":"https://pith.science/api/pith-number/KCG2TGZ7QLOP7SH3SZE7X24JRC/events.json","paper":"https://pith.science/paper/KCG2TGZ7"},"agent_actions":{"view_html":"https://pith.science/pith/KCG2TGZ7QLOP7SH3SZE7X24JRC","download_json":"https://pith.science/pith/KCG2TGZ7QLOP7SH3SZE7X24JRC.json","view_paper":"https://pith.science/paper/KCG2TGZ7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.24171&json=true","fetch_graph":"https://pith.science/api/pith-number/KCG2TGZ7QLOP7SH3SZE7X24JRC/graph.json","fetch_events":"https://pith.science/api/pith-number/KCG2TGZ7QLOP7SH3SZE7X24JRC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KCG2TGZ7QLOP7SH3SZE7X24JRC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KCG2TGZ7QLOP7SH3SZE7X24JRC/action/storage_attestation","attest_author":"https://pith.science/pith/KCG2TGZ7QLOP7SH3SZE7X24JRC/action/author_attestation","sign_citation":"https://pith.science/pith/KCG2TGZ7QLOP7SH3SZE7X24JRC/action/citation_signature","submit_replication":"https://pith.science/pith/KCG2TGZ7QLOP7SH3SZE7X24JRC/action/replication_record"}},"created_at":"2026-05-26T01:02:50.439021+00:00","updated_at":"2026-05-26T01:02:50.439021+00:00"}