{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:HVF3LOTTC34HDWC2LTG2VQFFI3","short_pith_number":"pith:HVF3LOTT","canonical_record":{"source":{"id":"2305.01210","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SE","submitted_at":"2023-05-02T05:46:48Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"e6912ff5b6a9a8d99edc6c7fc3fed66c47a34217d52c0df0c24b74db00f741a2","abstract_canon_sha256":"04b221cba3aa676fae32679075e1337b1723cea41e1848f41ca3a87499c2be97"},"schema_version":"1.0"},"canonical_sha256":"3d4bb5ba7316f871d85a5ccdaac0a546e902c3c27765137d80ddee3bd3d8c681","source":{"kind":"arxiv","id":"2305.01210","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2305.01210","created_at":"2026-05-18T02:44:08Z"},{"alias_kind":"arxiv_version","alias_value":"2305.01210v3","created_at":"2026-05-18T02:44:08Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.01210","created_at":"2026-05-18T02:44:08Z"},{"alias_kind":"pith_short_12","alias_value":"HVF3LOTTC34H","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"HVF3LOTTC34HDWC2","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"HVF3LOTT","created_at":"2026-05-18T12:33:33Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:HVF3LOTTC34HDWC2LTG2VQFFI3","target":"record","payload":{"canonical_record":{"source":{"id":"2305.01210","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SE","submitted_at":"2023-05-02T05:46:48Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"e6912ff5b6a9a8d99edc6c7fc3fed66c47a34217d52c0df0c24b74db00f741a2","abstract_canon_sha256":"04b221cba3aa676fae32679075e1337b1723cea41e1848f41ca3a87499c2be97"},"schema_version":"1.0"},"canonical_sha256":"3d4bb5ba7316f871d85a5ccdaac0a546e902c3c27765137d80ddee3bd3d8c681","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:44:08.792857Z","signature_b64":"sA80ZV9EHavQzk/VVsV1KEVA6AkhGdoYoy5fFFI8OpRQ3tuDGmpJAxU79AM2DdBeKY/GWo6HA44LU6yawYhYBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3d4bb5ba7316f871d85a5ccdaac0a546e902c3c27765137d80ddee3bd3d8c681","last_reissued_at":"2026-05-18T02:44:08.792377Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:44:08.792377Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2305.01210","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T02:44:08Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5yys5jQfwf8BtLxHH08qgp6DX3EQ5u2tOQeAY/s4wvhikZLDfWKmn8w+u23Szihv66jk4rUmLrSwCJsNTo+iCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-20T23:04:50.475893Z"},"content_sha256":"e44fbb8b6258cb0ae78656f293a539aaa785ccc6db6b0191ed5648200b98c77d","schema_version":"1.0","event_id":"sha256:e44fbb8b6258cb0ae78656f293a539aaa785ccc6db6b0191ed5648200b98c77d"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:HVF3LOTTC34HDWC2LTG2VQFFI3","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Augmenting HumanEval with 80 times more test cases reveals that LLM-generated code contains substantially more functional errors than prior benchmarks detected.","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.SE","authors_text":"Chunqiu Steven Xia, Jiawei Liu, Lingming Zhang, Yuyao Wang","submitted_at":"2023-05-02T05:46:48Z","abstract_excerpt":"Program synthesis has been long studied with recent approaches focused on directly using the power of Large Language Models (LLMs) to generate code. Programming benchmarks, with curated synthesis problems and test-cases, are used to measure the performance of various LLMs on code synthesis. However, these test-cases can be limited in both quantity and quality for fully assessing the functional correctness of the generated code. Such limitation in the existing benchmarks begs the following question: In the era of LLMs, is the code generated really correct? To answer this, we propose EvalPlus --"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our extensive evaluation across 26 popular LLMs demonstrates that HumanEval+ is able to catch significant amounts of previously undetected wrong code synthesized by LLMs, reducing the pass@k by up-to 19.3-28.9%.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The automatically generated test cases are functionally correct and do not introduce false failures or miss important edge cases in the code under test.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"EvalPlus augments HumanEval with 80x more tests via LLM and mutation strategies, exposing up to 28.9% more incorrect LLM-generated code and reversing some model performance rankings.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Augmenting HumanEval with 80 times more test cases reveals that LLM-generated code contains substantially more functional errors than prior benchmarks detected.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f4fb22ccd58e27830b15f68b44b2220c87919a62ae689ce9c9e453737b4bc643"},"source":{"id":"2305.01210","kind":"arxiv","version":3},"verdict":{"id":"8a32ed04-ba68-4560-bbd4-28249b6fa9a9","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T01:58:11.769884Z","strongest_claim":"Our extensive evaluation across 26 popular LLMs demonstrates that HumanEval+ is able to catch significant amounts of previously undetected wrong code synthesized by LLMs, reducing the pass@k by up-to 19.3-28.9%.","one_line_summary":"EvalPlus augments HumanEval with 80x more tests via LLM and mutation strategies, exposing up to 28.9% more incorrect LLM-generated code and reversing some model performance rankings.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The automatically generated test cases are functionally correct and do not introduce false failures or miss important edge cases in the code under test.","pith_extraction_headline":"Augmenting HumanEval with 80 times more test cases reveals that LLM-generated code contains substantially more functional errors than prior benchmarks detected."},"references":{"count":76,"sample":[{"doi":"","year":2022,"title":"T. Ahmed and P. Devanbu. Few-shot training llms for project-specific code-summarization. In 37th IEEE/ACM International Conference on Automated Software Engineering, pages 1–5, 2022","work_id":"1bb08f1e-203e-4f1f-a10c-f01969ecedb9","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Santacoder: don’t reach for the stars! arXiv preprint arXiv:2301.03988","work_id":"bf393c50-a11b-4a0f-8513-52428ede71f7","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"J. Austin, A. Odena, M. Nye, M. Bosma, H. Michalewski, D. Dohan, E. Jiang, C. Cai, M. Terry, Q. Le, and C. Sutton. Program synthesis with large language models, 2021","work_id":"214a9f12-b122-4033-8e0c-c6414d4ff463","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"S. Bang, S. Nam, I. Chun, H. Y . Jhoo, and J. Lee. Smt-based translation validation for machine learning compiler. In Computer Aided Verification: 34th International Conference, CAV 2022, Haifa, Israe","work_id":"eaf2685d-5d0e-4925-8b9a-089a989d9bae","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"S. Black, L. Gao, P. Wang, C. Leahy, and S. Biderman. GPT-Neo: Large Scale Autoregressive Language Modeling with Mesh-Tensorflow, Mar. 2021. If you use this software, please cite it using these metada","work_id":"7300281c-8472-4002-9d18-dd583a8986d3","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":76,"snapshot_sha256":"dc74bba5f761852b703d8e587e8f41fa5e683a46d8237897d95a747c48fd1e7d","internal_anchors":6},"formal_canon":{"evidence_count":2,"snapshot_sha256":"3d720ce297ebb3c900608d8650b78878727ef351a1186990a4cf63b72069ccf0"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"8a32ed04-ba68-4560-bbd4-28249b6fa9a9"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T02:44:08Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"gucJmMI3Zu7r5BFK/9Caesp0Fe97yX64S8bQlEqpwBMAIoogIBw96BwEuk7wjQdm+f+5xyxH6LF1b4p4wvVsCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-20T23:04:50.476654Z"},"content_sha256":"18ea91709155343faa3705d6ca28f79c0f295a67fc60f470d8db9a8547e83041","schema_version":"1.0","event_id":"sha256:18ea91709155343faa3705d6ca28f79c0f295a67fc60f470d8db9a8547e83041"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/HVF3LOTTC34HDWC2LTG2VQFFI3/bundle.json","state_url":"https://pith.science/pith/HVF3LOTTC34HDWC2LTG2VQFFI3/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/HVF3LOTTC34HDWC2LTG2VQFFI3/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-20T23:04:50Z","links":{"resolver":"https://pith.science/pith/HVF3LOTTC34HDWC2LTG2VQFFI3","bundle":"https://pith.science/pith/HVF3LOTTC34HDWC2LTG2VQFFI3/bundle.json","state":"https://pith.science/pith/HVF3LOTTC34HDWC2LTG2VQFFI3/state.json","well_known_bundle":"https://pith.science/.well-known/pith/HVF3LOTTC34HDWC2LTG2VQFFI3/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:HVF3LOTTC34HDWC2LTG2VQFFI3","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"04b221cba3aa676fae32679075e1337b1723cea41e1848f41ca3a87499c2be97","cross_cats_sorted":["cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SE","submitted_at":"2023-05-02T05:46:48Z","title_canon_sha256":"e6912ff5b6a9a8d99edc6c7fc3fed66c47a34217d52c0df0c24b74db00f741a2"},"schema_version":"1.0","source":{"id":"2305.01210","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2305.01210","created_at":"2026-05-18T02:44:08Z"},{"alias_kind":"arxiv_version","alias_value":"2305.01210v3","created_at":"2026-05-18T02:44:08Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.01210","created_at":"2026-05-18T02:44:08Z"},{"alias_kind":"pith_short_12","alias_value":"HVF3LOTTC34H","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"HVF3LOTTC34HDWC2","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"HVF3LOTT","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:18ea91709155343faa3705d6ca28f79c0f295a67fc60f470d8db9a8547e83041","target":"graph","created_at":"2026-05-18T02:44:08Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Our extensive evaluation across 26 popular LLMs demonstrates that HumanEval+ is able to catch significant amounts of previously undetected wrong code synthesized by LLMs, reducing the pass@k by up-to 19.3-28.9%."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The automatically generated test cases are functionally correct and do not introduce false failures or miss important edge cases in the code under test."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"EvalPlus augments HumanEval with 80x more tests via LLM and mutation strategies, exposing up to 28.9% more incorrect LLM-generated code and reversing some model performance rankings."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Augmenting HumanEval with 80 times more test cases reveals that LLM-generated code contains substantially more functional errors than prior benchmarks detected."}],"snapshot_sha256":"f4fb22ccd58e27830b15f68b44b2220c87919a62ae689ce9c9e453737b4bc643"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"3d720ce297ebb3c900608d8650b78878727ef351a1186990a4cf63b72069ccf0"},"paper":{"abstract_excerpt":"Program synthesis has been long studied with recent approaches focused on directly using the power of Large Language Models (LLMs) to generate code. Programming benchmarks, with curated synthesis problems and test-cases, are used to measure the performance of various LLMs on code synthesis. However, these test-cases can be limited in both quantity and quality for fully assessing the functional correctness of the generated code. Such limitation in the existing benchmarks begs the following question: In the era of LLMs, is the code generated really correct? To answer this, we propose EvalPlus --","authors_text":"Chunqiu Steven Xia, Jiawei Liu, Lingming Zhang, Yuyao Wang","cross_cats":["cs.CL","cs.LG"],"headline":"Augmenting HumanEval with 80 times more test cases reveals that LLM-generated code contains substantially more functional errors than prior benchmarks detected.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SE","submitted_at":"2023-05-02T05:46:48Z","title":"Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation"},"references":{"count":76,"internal_anchors":6,"resolved_work":76,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"T. Ahmed and P. Devanbu. Few-shot training llms for project-specific code-summarization. In 37th IEEE/ACM International Conference on Automated Software Engineering, pages 1–5, 2022","work_id":"1bb08f1e-203e-4f1f-a10c-f01969ecedb9","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Santacoder: don’t reach for the stars! arXiv preprint arXiv:2301.03988","work_id":"bf393c50-a11b-4a0f-8513-52428ede71f7","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"J. Austin, A. Odena, M. Nye, M. Bosma, H. Michalewski, D. Dohan, E. Jiang, C. Cai, M. Terry, Q. Le, and C. Sutton. Program synthesis with large language models, 2021","work_id":"214a9f12-b122-4033-8e0c-c6414d4ff463","year":2021},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"S. Bang, S. Nam, I. Chun, H. Y . Jhoo, and J. Lee. Smt-based translation validation for machine learning compiler. In Computer Aided Verification: 34th International Conference, CAV 2022, Haifa, Israe","work_id":"eaf2685d-5d0e-4925-8b9a-089a989d9bae","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"S. Black, L. Gao, P. Wang, C. Leahy, and S. Biderman. GPT-Neo: Large Scale Autoregressive Language Modeling with Mesh-Tensorflow, Mar. 2021. If you use this software, please cite it using these metada","work_id":"7300281c-8472-4002-9d18-dd583a8986d3","year":2021}],"snapshot_sha256":"dc74bba5f761852b703d8e587e8f41fa5e683a46d8237897d95a747c48fd1e7d"},"source":{"id":"2305.01210","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-14T01:58:11.769884Z","id":"8a32ed04-ba68-4560-bbd4-28249b6fa9a9","model_set":{"reader":"grok-4.3"},"one_line_summary":"EvalPlus augments HumanEval with 80x more tests via LLM and mutation strategies, exposing up to 28.9% more incorrect LLM-generated code and reversing some model performance rankings.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Augmenting HumanEval with 80 times more test cases reveals that LLM-generated code contains substantially more functional errors than prior benchmarks detected.","strongest_claim":"Our extensive evaluation across 26 popular LLMs demonstrates that HumanEval+ is able to catch significant amounts of previously undetected wrong code synthesized by LLMs, reducing the pass@k by up-to 19.3-28.9%.","weakest_assumption":"The automatically generated test cases are functionally correct and do not introduce false failures or miss important edge cases in the code under test."}},"verdict_id":"8a32ed04-ba68-4560-bbd4-28249b6fa9a9"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:e44fbb8b6258cb0ae78656f293a539aaa785ccc6db6b0191ed5648200b98c77d","target":"record","created_at":"2026-05-18T02:44:08Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"04b221cba3aa676fae32679075e1337b1723cea41e1848f41ca3a87499c2be97","cross_cats_sorted":["cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SE","submitted_at":"2023-05-02T05:46:48Z","title_canon_sha256":"e6912ff5b6a9a8d99edc6c7fc3fed66c47a34217d52c0df0c24b74db00f741a2"},"schema_version":"1.0","source":{"id":"2305.01210","kind":"arxiv","version":3}},"canonical_sha256":"3d4bb5ba7316f871d85a5ccdaac0a546e902c3c27765137d80ddee3bd3d8c681","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3d4bb5ba7316f871d85a5ccdaac0a546e902c3c27765137d80ddee3bd3d8c681","first_computed_at":"2026-05-18T02:44:08.792377Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T02:44:08.792377Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"sA80ZV9EHavQzk/VVsV1KEVA6AkhGdoYoy5fFFI8OpRQ3tuDGmpJAxU79AM2DdBeKY/GWo6HA44LU6yawYhYBg==","signature_status":"signed_v1","signed_at":"2026-05-18T02:44:08.792857Z","signed_message":"canonical_sha256_bytes"},"source_id":"2305.01210","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e44fbb8b6258cb0ae78656f293a539aaa785ccc6db6b0191ed5648200b98c77d","sha256:18ea91709155343faa3705d6ca28f79c0f295a67fc60f470d8db9a8547e83041"],"state_sha256":"6e60edde92fcf5262df09a475f3d18371e5837dcaf799fc22d32d91b49b4fe57"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0nqy6PrCgPorLE6/PXxQMXTQFAkSJfXb54ZNFtAzZmfzFayR5asAR9aw/ftxwHD3LhDZC+6t0q1Km6elOsscBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-20T23:04:50.480107Z","bundle_sha256":"c594d2e3d70cc9bc9c420db528a045d84970513eb1a6a9e67c3c609b4ae0622d"}}