{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:RM3CKGKHBVEIUIDUPN6YE7V2LN","short_pith_number":"pith:RM3CKGKH","schema_version":"1.0","canonical_sha256":"8b362519470d488a20747b7d827eba5b4e669993c42ee9af482450fefde3dc01","source":{"kind":"arxiv","id":"2607.02294","version":1},"attestation_state":"computed","paper":{"title":"Coding Agents Are Guessing: Measuring Action-Boundary Violations in Underspecified DevOps Instructions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.SE","authors_text":"Congying Xu, Shing-Chi Cheung, Shuai Wang, Yudong Gao, Zekai Zhang, Zimo Ji, Zongjie Li","submitted_at":"2026-07-02T15:11:39Z","abstract_excerpt":"LLM coding agents are increasingly deployed to act autonomously on real production infrastructure. They execute shell commands, modify repositories, and call operational APIs. However, completing a task is not sufficient for safety. A wrong action can cause severe consequences. Existing agent benchmarks largely emphasize task completion, leaving open how agents behave under benign but underspecified instructions.\n  We present UnderSpecBench, a benchmark for measuring action-boundary violations in coding agents (i.e., Claude Code, Codex, and OpenCode) on DevOps tasks. UnderSpecBench includes 69"},"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":"2607.02294","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SE","submitted_at":"2026-07-02T15:11:39Z","cross_cats_sorted":[],"title_canon_sha256":"52203ee79402c088ad4a26ee1e80e37a0227261d213fe67f9a91bfb86cf354a1","abstract_canon_sha256":"3852338a62d516536245bc9a3b88c79f46a10b14bb50b889627aeaabbbfd4c36"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-03T01:17:47.766775Z","signature_b64":"DtHxm+KzonoKrePd0mBajgv98GxOqbCCjE5PQay+49ce6XEikcnc3ij3gy+2+IMZOTAyncpKN3rG476KncpaDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8b362519470d488a20747b7d827eba5b4e669993c42ee9af482450fefde3dc01","last_reissued_at":"2026-07-03T01:17:47.766343Z","signature_status":"signed_v1","first_computed_at":"2026-07-03T01:17:47.766343Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Coding Agents Are Guessing: Measuring Action-Boundary Violations in Underspecified DevOps Instructions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.SE","authors_text":"Congying Xu, Shing-Chi Cheung, Shuai Wang, Yudong Gao, Zekai Zhang, Zimo Ji, Zongjie Li","submitted_at":"2026-07-02T15:11:39Z","abstract_excerpt":"LLM coding agents are increasingly deployed to act autonomously on real production infrastructure. They execute shell commands, modify repositories, and call operational APIs. However, completing a task is not sufficient for safety. A wrong action can cause severe consequences. Existing agent benchmarks largely emphasize task completion, leaving open how agents behave under benign but underspecified instructions.\n  We present UnderSpecBench, a benchmark for measuring action-boundary violations in coding agents (i.e., Claude Code, Codex, and OpenCode) on DevOps tasks. UnderSpecBench includes 69"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.02294","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/2607.02294/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":"2607.02294","created_at":"2026-07-03T01:17:47.766419+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.02294v1","created_at":"2026-07-03T01:17:47.766419+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.02294","created_at":"2026-07-03T01:17:47.766419+00:00"},{"alias_kind":"pith_short_12","alias_value":"RM3CKGKHBVEI","created_at":"2026-07-03T01:17:47.766419+00:00"},{"alias_kind":"pith_short_16","alias_value":"RM3CKGKHBVEIUIDU","created_at":"2026-07-03T01:17:47.766419+00:00"},{"alias_kind":"pith_short_8","alias_value":"RM3CKGKH","created_at":"2026-07-03T01:17:47.766419+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/RM3CKGKHBVEIUIDUPN6YE7V2LN","json":"https://pith.science/pith/RM3CKGKHBVEIUIDUPN6YE7V2LN.json","graph_json":"https://pith.science/api/pith-number/RM3CKGKHBVEIUIDUPN6YE7V2LN/graph.json","events_json":"https://pith.science/api/pith-number/RM3CKGKHBVEIUIDUPN6YE7V2LN/events.json","paper":"https://pith.science/paper/RM3CKGKH"},"agent_actions":{"view_html":"https://pith.science/pith/RM3CKGKHBVEIUIDUPN6YE7V2LN","download_json":"https://pith.science/pith/RM3CKGKHBVEIUIDUPN6YE7V2LN.json","view_paper":"https://pith.science/paper/RM3CKGKH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.02294&json=true","fetch_graph":"https://pith.science/api/pith-number/RM3CKGKHBVEIUIDUPN6YE7V2LN/graph.json","fetch_events":"https://pith.science/api/pith-number/RM3CKGKHBVEIUIDUPN6YE7V2LN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RM3CKGKHBVEIUIDUPN6YE7V2LN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RM3CKGKHBVEIUIDUPN6YE7V2LN/action/storage_attestation","attest_author":"https://pith.science/pith/RM3CKGKHBVEIUIDUPN6YE7V2LN/action/author_attestation","sign_citation":"https://pith.science/pith/RM3CKGKHBVEIUIDUPN6YE7V2LN/action/citation_signature","submit_replication":"https://pith.science/pith/RM3CKGKHBVEIUIDUPN6YE7V2LN/action/replication_record"}},"created_at":"2026-07-03T01:17:47.766419+00:00","updated_at":"2026-07-03T01:17:47.766419+00:00"}