{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:X2A2OF45SKTQ4FDANK534QK3Y7","short_pith_number":"pith:X2A2OF45","schema_version":"1.0","canonical_sha256":"be81a7179d92a70e14606abbbe415bc7f196fe5ce8f7d2930a598d4135514c9d","source":{"kind":"arxiv","id":"2506.03332","version":1},"attestation_state":"computed","paper":{"title":"Helpful Agent Meets Deceptive Judge: Understanding Vulnerabilities in Agentic Workflows","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Jiayu Wang, Shafiq Joty, Xuan-Phi Nguyen, Yifei Ming, Zixuan Ke","submitted_at":"2025-06-03T19:26:23Z","abstract_excerpt":"Agentic workflows -- where multiple large language model (LLM) instances interact to solve tasks -- are increasingly built on feedback mechanisms, where one model evaluates and critiques another. Despite the promise of feedback-driven improvement, the stability of agentic workflows rests on the reliability of the judge. However, judges may hallucinate information, exhibit bias, or act adversarially -- introducing critical vulnerabilities into the workflow. In this work, we present a systematic analysis of agentic workflows under deceptive or misleading feedback. We introduce a two-dimensional "},"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":"2506.03332","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2025-06-03T19:26:23Z","cross_cats_sorted":[],"title_canon_sha256":"4e09a9057d5a2c6604806c1e7127d2416a5b02b52fac74121585147e8cde43d1","abstract_canon_sha256":"624bd441e43c53131d6fd49930c406c993da64e29ce0b49d43aeb168c0d53413"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:15:25.367694Z","signature_b64":"JM/DJ68y+ggVt1cdi7wjTPDxV3O7x1pX+Or64qUodphFWG1H0SPryUAdCPoJtR5hj/hYvKdWshBi1xj4TuiVAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"be81a7179d92a70e14606abbbe415bc7f196fe5ce8f7d2930a598d4135514c9d","last_reissued_at":"2026-07-05T11:15:25.367154Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:15:25.367154Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Helpful Agent Meets Deceptive Judge: Understanding Vulnerabilities in Agentic Workflows","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Jiayu Wang, Shafiq Joty, Xuan-Phi Nguyen, Yifei Ming, Zixuan Ke","submitted_at":"2025-06-03T19:26:23Z","abstract_excerpt":"Agentic workflows -- where multiple large language model (LLM) instances interact to solve tasks -- are increasingly built on feedback mechanisms, where one model evaluates and critiques another. Despite the promise of feedback-driven improvement, the stability of agentic workflows rests on the reliability of the judge. However, judges may hallucinate information, exhibit bias, or act adversarially -- introducing critical vulnerabilities into the workflow. In this work, we present a systematic analysis of agentic workflows under deceptive or misleading feedback. We introduce a two-dimensional "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.03332","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/2506.03332/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":"2506.03332","created_at":"2026-07-05T11:15:25.367220+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.03332v1","created_at":"2026-07-05T11:15:25.367220+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.03332","created_at":"2026-07-05T11:15:25.367220+00:00"},{"alias_kind":"pith_short_12","alias_value":"X2A2OF45SKTQ","created_at":"2026-07-05T11:15:25.367220+00:00"},{"alias_kind":"pith_short_16","alias_value":"X2A2OF45SKTQ4FDA","created_at":"2026-07-05T11:15:25.367220+00:00"},{"alias_kind":"pith_short_8","alias_value":"X2A2OF45","created_at":"2026-07-05T11:15:25.367220+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.22534","citing_title":"Why Are Agentic Pull Requests Merged or Rejected? An Empirical Study","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11003","citing_title":"The Authorization-Execution Gap Is a Major Safety and Security Problem in Open-World Agents","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/X2A2OF45SKTQ4FDANK534QK3Y7","json":"https://pith.science/pith/X2A2OF45SKTQ4FDANK534QK3Y7.json","graph_json":"https://pith.science/api/pith-number/X2A2OF45SKTQ4FDANK534QK3Y7/graph.json","events_json":"https://pith.science/api/pith-number/X2A2OF45SKTQ4FDANK534QK3Y7/events.json","paper":"https://pith.science/paper/X2A2OF45"},"agent_actions":{"view_html":"https://pith.science/pith/X2A2OF45SKTQ4FDANK534QK3Y7","download_json":"https://pith.science/pith/X2A2OF45SKTQ4FDANK534QK3Y7.json","view_paper":"https://pith.science/paper/X2A2OF45","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.03332&json=true","fetch_graph":"https://pith.science/api/pith-number/X2A2OF45SKTQ4FDANK534QK3Y7/graph.json","fetch_events":"https://pith.science/api/pith-number/X2A2OF45SKTQ4FDANK534QK3Y7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/X2A2OF45SKTQ4FDANK534QK3Y7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/X2A2OF45SKTQ4FDANK534QK3Y7/action/storage_attestation","attest_author":"https://pith.science/pith/X2A2OF45SKTQ4FDANK534QK3Y7/action/author_attestation","sign_citation":"https://pith.science/pith/X2A2OF45SKTQ4FDANK534QK3Y7/action/citation_signature","submit_replication":"https://pith.science/pith/X2A2OF45SKTQ4FDANK534QK3Y7/action/replication_record"}},"created_at":"2026-07-05T11:15:25.367220+00:00","updated_at":"2026-07-05T11:15:25.367220+00:00"}