{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:Z2IMSEUSJEJQYIFMHSC6X2TTPT","short_pith_number":"pith:Z2IMSEUS","schema_version":"1.0","canonical_sha256":"ce90c9129249130c20ac3c85ebea737cfbdbf7c0f3d5f99e92246d6c85d8bd7d","source":{"kind":"arxiv","id":"2605.26177","version":1},"attestation_state":"computed","paper":{"title":"RepoMirage: Probing Repository Context Reasoning in Code Agents with Perturbations","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.SE","authors_text":"Hang Su, Hanyu Li, Jun Zhu, Speed Zhu, Yichi Zhang, Yinpeng Dong","submitted_at":"2026-05-25T06:26:43Z","abstract_excerpt":"Code agents are currently having skillful performance on repository-level software engineering benchmarks, but it remains unclear whether success on end-to-end tasks such as issue resolution truly reflects repository context reasoning, the ability to identify the task-relevant information across multiple files and reason over the relations among them. To investigate this question, we introduce RepoMirage, a two-stage evaluation suite built on SWE-Bench Verified that adopts perturbation as a diagnostic tool to increase the demand for context reasoning by transforming how the repository is expos"},"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":"2605.26177","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.SE","submitted_at":"2026-05-25T06:26:43Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"db1fbcb606e0c37ad644ff7f19fdbb10f72468ed7be07c880d526e0ad5f6916c","abstract_canon_sha256":"8a98fe7f932f1a25af9c2d8b45ca6c20ddc7c6142056e4cc25c49b92cf88b6f0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T00:04:27.495433Z","signature_b64":"anYD+oGiq0zLCxTx7oc4s6awxHwY0vKwRAIMlUX7Dq3UspaC3sLzvJ+bSq9N5IRUfLLS8/LVfJIIykvqvs9OCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ce90c9129249130c20ac3c85ebea737cfbdbf7c0f3d5f99e92246d6c85d8bd7d","last_reissued_at":"2026-05-27T00:04:27.494638Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T00:04:27.494638Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RepoMirage: Probing Repository Context Reasoning in Code Agents with Perturbations","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.SE","authors_text":"Hang Su, Hanyu Li, Jun Zhu, Speed Zhu, Yichi Zhang, Yinpeng Dong","submitted_at":"2026-05-25T06:26:43Z","abstract_excerpt":"Code agents are currently having skillful performance on repository-level software engineering benchmarks, but it remains unclear whether success on end-to-end tasks such as issue resolution truly reflects repository context reasoning, the ability to identify the task-relevant information across multiple files and reason over the relations among them. To investigate this question, we introduce RepoMirage, a two-stage evaluation suite built on SWE-Bench Verified that adopts perturbation as a diagnostic tool to increase the demand for context reasoning by transforming how the repository is expos"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26177","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/2605.26177/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":"2605.26177","created_at":"2026-05-27T00:04:27.494760+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.26177v1","created_at":"2026-05-27T00:04:27.494760+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26177","created_at":"2026-05-27T00:04:27.494760+00:00"},{"alias_kind":"pith_short_12","alias_value":"Z2IMSEUSJEJQ","created_at":"2026-05-27T00:04:27.494760+00:00"},{"alias_kind":"pith_short_16","alias_value":"Z2IMSEUSJEJQYIFM","created_at":"2026-05-27T00:04:27.494760+00:00"},{"alias_kind":"pith_short_8","alias_value":"Z2IMSEUS","created_at":"2026-05-27T00:04:27.494760+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/Z2IMSEUSJEJQYIFMHSC6X2TTPT","json":"https://pith.science/pith/Z2IMSEUSJEJQYIFMHSC6X2TTPT.json","graph_json":"https://pith.science/api/pith-number/Z2IMSEUSJEJQYIFMHSC6X2TTPT/graph.json","events_json":"https://pith.science/api/pith-number/Z2IMSEUSJEJQYIFMHSC6X2TTPT/events.json","paper":"https://pith.science/paper/Z2IMSEUS"},"agent_actions":{"view_html":"https://pith.science/pith/Z2IMSEUSJEJQYIFMHSC6X2TTPT","download_json":"https://pith.science/pith/Z2IMSEUSJEJQYIFMHSC6X2TTPT.json","view_paper":"https://pith.science/paper/Z2IMSEUS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.26177&json=true","fetch_graph":"https://pith.science/api/pith-number/Z2IMSEUSJEJQYIFMHSC6X2TTPT/graph.json","fetch_events":"https://pith.science/api/pith-number/Z2IMSEUSJEJQYIFMHSC6X2TTPT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Z2IMSEUSJEJQYIFMHSC6X2TTPT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Z2IMSEUSJEJQYIFMHSC6X2TTPT/action/storage_attestation","attest_author":"https://pith.science/pith/Z2IMSEUSJEJQYIFMHSC6X2TTPT/action/author_attestation","sign_citation":"https://pith.science/pith/Z2IMSEUSJEJQYIFMHSC6X2TTPT/action/citation_signature","submit_replication":"https://pith.science/pith/Z2IMSEUSJEJQYIFMHSC6X2TTPT/action/replication_record"}},"created_at":"2026-05-27T00:04:27.494760+00:00","updated_at":"2026-05-27T00:04:27.494760+00:00"}