{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:WIBKBDV4KRKOFYE3MOVUGKOGTE","short_pith_number":"pith:WIBKBDV4","canonical_record":{"source":{"id":"2406.15877","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2024-06-22T15:52:04Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"224b55f50f07d308debafc307feffe7fc059e5057e530715828242735aa4cb43","abstract_canon_sha256":"e27ee9ec1d97a8e6fe8dc014c14fe2b3f77ac5e8a07139bc07cc8139bcea59fa"},"schema_version":"1.0"},"canonical_sha256":"b202a08ebc5454e2e09b63ab4329c6991b6ad92dac721646508438456ba6a097","source":{"kind":"arxiv","id":"2406.15877","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.15877","created_at":"2026-05-17T23:39:22Z"},{"alias_kind":"arxiv_version","alias_value":"2406.15877v4","created_at":"2026-05-17T23:39:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.15877","created_at":"2026-05-17T23:39:22Z"},{"alias_kind":"pith_short_12","alias_value":"WIBKBDV4KRKO","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"WIBKBDV4KRKOFYE3","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"WIBKBDV4","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:WIBKBDV4KRKOFYE3MOVUGKOGTE","target":"record","payload":{"canonical_record":{"source":{"id":"2406.15877","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2024-06-22T15:52:04Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"224b55f50f07d308debafc307feffe7fc059e5057e530715828242735aa4cb43","abstract_canon_sha256":"e27ee9ec1d97a8e6fe8dc014c14fe2b3f77ac5e8a07139bc07cc8139bcea59fa"},"schema_version":"1.0"},"canonical_sha256":"b202a08ebc5454e2e09b63ab4329c6991b6ad92dac721646508438456ba6a097","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:22.051431Z","signature_b64":"piOmrqWuxGTMDw1ECeYXiarGscWAUjM7J0AKJW/UN68oibYUQTz1hJhBg2RCRua8FyXwdQlOprHYtJes1KsmBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b202a08ebc5454e2e09b63ab4329c6991b6ad92dac721646508438456ba6a097","last_reissued_at":"2026-05-17T23:39:22.050812Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:22.050812Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2406.15877","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:22Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"f6QTk43ApiQ/f3rkHBUCh3penWWvuxAyOB++ABm3ANR/rkZgpmGYJ5VNQAMPSvlpNbDdfSOiUeCZH2R6oLuAAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-28T06:23:32.414494Z"},"content_sha256":"55b83d766c6e37637862110f52b294a7ba23674a195144222ed91269574f35c6","schema_version":"1.0","event_id":"sha256:55b83d766c6e37637862110f52b294a7ba23674a195144222ed91269574f35c6"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:WIBKBDV4KRKOFYE3MOVUGKOGTE","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"BigCodeBench: Benchmarking Code Generation with Diverse Function Calls and Complex Instructions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Large language models reach only up to 60 percent success on tasks requiring precise use of diverse function calls from many libraries, far below the 97 percent human level.","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.SE","authors_text":"Alex Gu, Armel Randy Zebaze, Binyuan Hui, Chen Gong, Daniel Fried, David Lo, Han Hu, Haolan Zhan, Harm de Vries, Imam Nur Bani Yusuf, Indraneil Paul, Jean Kaddour, Jenny Chim, Jiawei Liu, Junda He, Leandro Von Werra, Ming Xu, Minh Chien Vu, Naman Jain, Niklas Muennighoff, Prateek Yadav, Qian Liu, Ratnadira Widyasari, Simon Brunner, Terry Yue Zhuo, Thong Hoang, Wen-Ding Li, Wenhao Yu, Xiaoheng Hong, Xiaoning Du, Zhihan Zhang, Zhoujun Cheng, Zijian Wang","submitted_at":"2024-06-22T15:52:04Z","abstract_excerpt":"Task automation has been greatly empowered by the recent advances in Large Language Models (LLMs) via Python code, where the tasks ranging from software engineering development to general-purpose reasoning. While current benchmarks have shown that LLMs can solve tasks using programs like human developers, the majority of their evaluations are limited to short and self-contained algorithmic tasks or standalone function calls. Solving challenging and practical tasks requires the capability of utilizing diverse function calls as tools to efficiently implement functionalities like data analysis an"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our extensive evaluation of 60 LLMs shows that LLMs are not yet capable of following complex instructions to use function calls precisely, with scores up to 60%, significantly lower than the human performance of 97%.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The 1,140 tasks and their test cases accurately represent the challenges of real-world practical coding that requires diverse function calls from many libraries.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"BigCodeBench shows LLMs achieve at most 60% on 1,140 tasks needing diverse function calls and complex instructions, compared to 97% human performance.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Large language models reach only up to 60 percent success on tasks requiring precise use of diverse function calls from many libraries, far below the 97 percent human level.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"927ec8f0b499503b33e0a6a86b0154413e22af9972fe4aa6d0d3a63ffb8b9665"},"source":{"id":"2406.15877","kind":"arxiv","version":4},"verdict":{"id":"6085f7d0-88fd-4213-8284-11b6abe7afd3","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:24:25.315333Z","strongest_claim":"Our extensive evaluation of 60 LLMs shows that LLMs are not yet capable of following complex instructions to use function calls precisely, with scores up to 60%, significantly lower than the human performance of 97%.","one_line_summary":"BigCodeBench shows LLMs achieve at most 60% on 1,140 tasks needing diverse function calls and complex instructions, compared to 97% human performance.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The 1,140 tasks and their test cases accurately represent the challenges of real-world practical coding that requires diverse function calls from many libraries.","pith_extraction_headline":"Large language models reach only up to 60 percent success on tasks requiring precise use of diverse function calls from many libraries, far below the 97 percent human level."},"references":{"count":19,"sample":[{"doi":"","year":2021,"title":"Later, OpenCodeInterpreter (Zheng et al., 2024b) developed a multi-turn instruction dataset and achieved better coding performance","work_id":"ea964ef2-d29e-49dd-aab2-69434b7d8f7b","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2014,"title":"How well do the models generalize to the unseen tools and tasks?","work_id":"18398025-a81d-44fa-8707-621dc88a1aa8","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"This means when you see the function stub and docstring, you should be able to implement ← - with exactly the same functionality with the given function body","work_id":"4965fb86-54d4-44a6-81e5-b89a89d446c1","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"write blackbox unit tests to ensure the functional correctness of the given function. You should also ← - make the function easy to test. ### Step1:Check Library Imports #### Import Statement - Remove","work_id":"a3af4011-ef02-43d1-aaaf-4f9642f9e4c4","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"This prevents the ← - user from inferring the function’s purpose based on its name","work_id":"e9ed80fc-391f-47e1-af53-81ff7a09c017","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":19,"snapshot_sha256":"5dd26057d36e8e4d839894bc92b51d285e7b41e0eb5cd63f5770456f106a610b","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b9fa4dbed969ce0b8dc4549c75f85616de84b6157f197810e087783e1c31421d"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"6085f7d0-88fd-4213-8284-11b6abe7afd3"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:22Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"nbKVrpT5oVWBemgGIv66VBD6kXN+8hP1mw+jruCQrtB+BOqqbrXPyorCTcV+9gsqhtR6P1jr3iwb9i5A+TQeDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-28T06:23:32.415174Z"},"content_sha256":"9a97d64f4732d3b324d5bffdf7b24170fa1608a7fdcd3646148692e05013c54d","schema_version":"1.0","event_id":"sha256:9a97d64f4732d3b324d5bffdf7b24170fa1608a7fdcd3646148692e05013c54d"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/WIBKBDV4KRKOFYE3MOVUGKOGTE/bundle.json","state_url":"https://pith.science/pith/WIBKBDV4KRKOFYE3MOVUGKOGTE/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/WIBKBDV4KRKOFYE3MOVUGKOGTE/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":"2026-05-28T06:23:32Z","links":{"resolver":"https://pith.science/pith/WIBKBDV4KRKOFYE3MOVUGKOGTE","bundle":"https://pith.science/pith/WIBKBDV4KRKOFYE3MOVUGKOGTE/bundle.json","state":"https://pith.science/pith/WIBKBDV4KRKOFYE3MOVUGKOGTE/state.json","well_known_bundle":"https://pith.science/.well-known/pith/WIBKBDV4KRKOFYE3MOVUGKOGTE/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:WIBKBDV4KRKOFYE3MOVUGKOGTE","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":"e27ee9ec1d97a8e6fe8dc014c14fe2b3f77ac5e8a07139bc07cc8139bcea59fa","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2024-06-22T15:52:04Z","title_canon_sha256":"224b55f50f07d308debafc307feffe7fc059e5057e530715828242735aa4cb43"},"schema_version":"1.0","source":{"id":"2406.15877","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.15877","created_at":"2026-05-17T23:39:22Z"},{"alias_kind":"arxiv_version","alias_value":"2406.15877v4","created_at":"2026-05-17T23:39:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.15877","created_at":"2026-05-17T23:39:22Z"},{"alias_kind":"pith_short_12","alias_value":"WIBKBDV4KRKO","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"WIBKBDV4KRKOFYE3","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"WIBKBDV4","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:9a97d64f4732d3b324d5bffdf7b24170fa1608a7fdcd3646148692e05013c54d","target":"graph","created_at":"2026-05-17T23:39:22Z","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 of 60 LLMs shows that LLMs are not yet capable of following complex instructions to use function calls precisely, with scores up to 60%, significantly lower than the human performance of 97%."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The 1,140 tasks and their test cases accurately represent the challenges of real-world practical coding that requires diverse function calls from many libraries."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"BigCodeBench shows LLMs achieve at most 60% on 1,140 tasks needing diverse function calls and complex instructions, compared to 97% human performance."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Large language models reach only up to 60 percent success on tasks requiring precise use of diverse function calls from many libraries, far below the 97 percent human level."}],"snapshot_sha256":"927ec8f0b499503b33e0a6a86b0154413e22af9972fe4aa6d0d3a63ffb8b9665"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b9fa4dbed969ce0b8dc4549c75f85616de84b6157f197810e087783e1c31421d"},"paper":{"abstract_excerpt":"Task automation has been greatly empowered by the recent advances in Large Language Models (LLMs) via Python code, where the tasks ranging from software engineering development to general-purpose reasoning. While current benchmarks have shown that LLMs can solve tasks using programs like human developers, the majority of their evaluations are limited to short and self-contained algorithmic tasks or standalone function calls. Solving challenging and practical tasks requires the capability of utilizing diverse function calls as tools to efficiently implement functionalities like data analysis an","authors_text":"Alex Gu, Armel Randy Zebaze, Binyuan Hui, Chen Gong, Daniel Fried, David Lo, Han Hu, Haolan Zhan, Harm de Vries, Imam Nur Bani Yusuf, Indraneil Paul, Jean Kaddour, Jenny Chim, Jiawei Liu, Junda He, Leandro Von Werra, Ming Xu, Minh Chien Vu, Naman Jain, Niklas Muennighoff, Prateek Yadav, Qian Liu, Ratnadira Widyasari, Simon Brunner, Terry Yue Zhuo, Thong Hoang, Wen-Ding Li, Wenhao Yu, Xiaoheng Hong, Xiaoning Du, Zhihan Zhang, Zhoujun Cheng, Zijian Wang","cross_cats":["cs.AI","cs.CL"],"headline":"Large language models reach only up to 60 percent success on tasks requiring precise use of diverse function calls from many libraries, far below the 97 percent human level.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2024-06-22T15:52:04Z","title":"BigCodeBench: Benchmarking Code Generation with Diverse Function Calls and Complex Instructions"},"references":{"count":19,"internal_anchors":0,"resolved_work":19,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Later, OpenCodeInterpreter (Zheng et al., 2024b) developed a multi-turn instruction dataset and achieved better coding performance","work_id":"ea964ef2-d29e-49dd-aab2-69434b7d8f7b","year":2021},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"How well do the models generalize to the unseen tools and tasks?","work_id":"18398025-a81d-44fa-8707-621dc88a1aa8","year":2014},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"This means when you see the function stub and docstring, you should be able to implement ← - with exactly the same functionality with the given function body","work_id":"4965fb86-54d4-44a6-81e5-b89a89d446c1","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"write blackbox unit tests to ensure the functional correctness of the given function. You should also ← - make the function easy to test. ### Step1:Check Library Imports #### Import Statement - Remove","work_id":"a3af4011-ef02-43d1-aaaf-4f9642f9e4c4","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"This prevents the ← - user from inferring the function’s purpose based on its name","work_id":"e9ed80fc-391f-47e1-af53-81ff7a09c017","year":null}],"snapshot_sha256":"5dd26057d36e8e4d839894bc92b51d285e7b41e0eb5cd63f5770456f106a610b"},"source":{"id":"2406.15877","kind":"arxiv","version":4},"verdict":{"created_at":"2026-05-14T19:24:25.315333Z","id":"6085f7d0-88fd-4213-8284-11b6abe7afd3","model_set":{"reader":"grok-4.3"},"one_line_summary":"BigCodeBench shows LLMs achieve at most 60% on 1,140 tasks needing diverse function calls and complex instructions, compared to 97% human performance.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Large language models reach only up to 60 percent success on tasks requiring precise use of diverse function calls from many libraries, far below the 97 percent human level.","strongest_claim":"Our extensive evaluation of 60 LLMs shows that LLMs are not yet capable of following complex instructions to use function calls precisely, with scores up to 60%, significantly lower than the human performance of 97%.","weakest_assumption":"The 1,140 tasks and their test cases accurately represent the challenges of real-world practical coding that requires diverse function calls from many libraries."}},"verdict_id":"6085f7d0-88fd-4213-8284-11b6abe7afd3"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:55b83d766c6e37637862110f52b294a7ba23674a195144222ed91269574f35c6","target":"record","created_at":"2026-05-17T23:39:22Z","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":"e27ee9ec1d97a8e6fe8dc014c14fe2b3f77ac5e8a07139bc07cc8139bcea59fa","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2024-06-22T15:52:04Z","title_canon_sha256":"224b55f50f07d308debafc307feffe7fc059e5057e530715828242735aa4cb43"},"schema_version":"1.0","source":{"id":"2406.15877","kind":"arxiv","version":4}},"canonical_sha256":"b202a08ebc5454e2e09b63ab4329c6991b6ad92dac721646508438456ba6a097","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b202a08ebc5454e2e09b63ab4329c6991b6ad92dac721646508438456ba6a097","first_computed_at":"2026-05-17T23:39:22.050812Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:22.050812Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"piOmrqWuxGTMDw1ECeYXiarGscWAUjM7J0AKJW/UN68oibYUQTz1hJhBg2RCRua8FyXwdQlOprHYtJes1KsmBQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:22.051431Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.15877","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:55b83d766c6e37637862110f52b294a7ba23674a195144222ed91269574f35c6","sha256:9a97d64f4732d3b324d5bffdf7b24170fa1608a7fdcd3646148692e05013c54d"],"state_sha256":"b455e601ee6135790da102938ee52bf52c0c13d5e619b10ae9f71bf71e8a03f8"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"v5lSPyj/CnSb0/hy9YPygiGCM0k+z+p93dE3EcsMzRSWzMjm/odYgcZCdeYtI20mA+3o3Mspgm690Puu/QuSAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-28T06:23:32.417855Z","bundle_sha256":"15d5bad45870a63221f099261fe4e7dd4f1adb03dd4c4a282d3ae5a0fe59e536"}}