{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:WPYOEUJUCS3Y3LBJVJUDLPDCP7","short_pith_number":"pith:WPYOEUJU","canonical_record":{"source":{"id":"2605.15206","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-01T14:45:37Z","cross_cats_sorted":["cs.AI","cs.DC"],"title_canon_sha256":"3567ce10ea6e5489677de443cb3d2df6c721c84cf4302d53e421faa611400dd8","abstract_canon_sha256":"6dd79ec1b2806626a527bc7c0a30333a3f874b06d1dad3060f1207494846640c"},"schema_version":"1.0"},"canonical_sha256":"b3f0e2513414b78dac29aa6835bc627fff91fc66a8cd66f2598f3798e949479b","source":{"kind":"arxiv","id":"2605.15206","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.15206","created_at":"2026-05-20T00:00:46Z"},{"alias_kind":"arxiv_version","alias_value":"2605.15206v1","created_at":"2026-05-20T00:00:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15206","created_at":"2026-05-20T00:00:46Z"},{"alias_kind":"pith_short_12","alias_value":"WPYOEUJUCS3Y","created_at":"2026-05-20T00:00:46Z"},{"alias_kind":"pith_short_16","alias_value":"WPYOEUJUCS3Y3LBJ","created_at":"2026-05-20T00:00:46Z"},{"alias_kind":"pith_short_8","alias_value":"WPYOEUJU","created_at":"2026-05-20T00:00:46Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:WPYOEUJUCS3Y3LBJVJUDLPDCP7","target":"record","payload":{"canonical_record":{"source":{"id":"2605.15206","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-01T14:45:37Z","cross_cats_sorted":["cs.AI","cs.DC"],"title_canon_sha256":"3567ce10ea6e5489677de443cb3d2df6c721c84cf4302d53e421faa611400dd8","abstract_canon_sha256":"6dd79ec1b2806626a527bc7c0a30333a3f874b06d1dad3060f1207494846640c"},"schema_version":"1.0"},"canonical_sha256":"b3f0e2513414b78dac29aa6835bc627fff91fc66a8cd66f2598f3798e949479b","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:46.052328Z","signature_b64":"qFPaudcY9b6W0Z7L5pwoorP9wDSkw+2+L6NSfHsV1sb/O3cA+3riBo3m27vYIjUpVbHJdR3juH/qTrZscYjoBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b3f0e2513414b78dac29aa6835bc627fff91fc66a8cd66f2598f3798e949479b","last_reissued_at":"2026-05-20T00:00:46.051207Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:46.051207Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.15206","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:00:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"wU3Z3mulV8asUvLPlazilLg5lvcWF7XrshJx5cpLjppq1YZoPaVwrcfY7RkxGOAubskmuCrNSbtgx/klUStzBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-21T20:14:10.566275Z"},"content_sha256":"f4dbae77e7c0654ba11dbf69b9dd5b928732b6e86518759211f378d65cfa66ca","schema_version":"1.0","event_id":"sha256:f4dbae77e7c0654ba11dbf69b9dd5b928732b6e86518759211f378d65cfa66ca"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:WPYOEUJUCS3Y3LBJVJUDLPDCP7","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"AgentStop: Terminating Local AI Agents Early to Save Energy in Consumer Devices","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Local LLM agents can stop themselves early using token log probabilities to cut wasted energy by 15-20% on consumer devices.","cross_cats":["cs.AI","cs.DC"],"primary_cat":"cs.LG","authors_text":"Ali Shahin Shamsabadi, Dzung Pham, Hamed Haddadi, Kleomenis Katevas","submitted_at":"2026-05-01T14:45:37Z","abstract_excerpt":"Autonomous agents powered by large language models (LLMs) are increasingly used to automate complex, multi-step tasks such as coding or web-based question answering. While remote, cloud-based agents offer scalability and ease of deployment, they raise privacy concerns, depend on network connectivity, and incur recurring API costs. Deploying agents locally on user devices mitigates these issues by preserving data privacy and eliminating usage-based fees. However, agentic workflows are far more resource-intensive than typical LLM interactions. Iterative reasoning, tool use, and failure retries s"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Leveraging low-cost execution signals such as token-level log probabilities, AgentStop can reduce wasted energy by 15-20% with minimal impact on task performance (<5% utility drop) for challenging web-based question answering and coding benchmarks.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That token-level log probabilities and similar low-cost signals during execution are sufficiently predictive of ultimate task failure to allow early termination without introducing many harmful false positives that degrade overall utility.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"AgentStop uses execution signals to early-terminate failing local LLM agent trajectories, cutting energy use 15-20% with minimal utility loss.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Local LLM agents can stop themselves early using token log probabilities to cut wasted energy by 15-20% on consumer devices.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"6ca0f6f4919f6305ace5ff13d09d3f360730dbc85e8ef0a690c0f9c51f143b0f"},"source":{"id":"2605.15206","kind":"arxiv","version":1},"verdict":{"id":"7cee7dae-c3aa-4bb2-8828-8e55c55f8419","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T16:31:23.737876Z","strongest_claim":"Leveraging low-cost execution signals such as token-level log probabilities, AgentStop can reduce wasted energy by 15-20% with minimal impact on task performance (<5% utility drop) for challenging web-based question answering and coding benchmarks.","one_line_summary":"AgentStop uses execution signals to early-terminate failing local LLM agent trajectories, cutting energy use 15-20% with minimal utility loss.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That token-level log probabilities and similar low-cost signals during execution are sufficiently predictive of ultimate task failure to allow early termination without introducing many harmful false positives that degrade overall utility.","pith_extraction_headline":"Local LLM agents can stop themselves early using token log probabilities to cut wasted energy by 15-20% on consumer devices."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.15206/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_compliance","ran_at":"2026-05-19T17:55:02.416062Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"d48aa00c2436c4a53e4cd64b4ac110f907cf88eb3cac0cd39630bf08002bd741"},"references":{"count":58,"sample":[{"doi":"","year":2026,"title":"Anthropic. 2026. Claude API Pricing. https://platform.claude.com/docs/en/about- claude/pricing. Accessed: 2026-02-28","work_id":"5f4a3af6-3677-4c19-bf4a-9f92a80d5aac","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Small Language Models are the Future of Agentic AI","work_id":"ba0f0305-4a51-48fd-a13f-201439a18f9e","ref_index":2,"cited_arxiv_id":"2506.02153","is_internal_anchor":true},{"doi":"","year":2016,"title":"Why Should I Trust You?","work_id":"9e35d1c4-f9d4-46bb-9508-e0c864a2560a","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Measuring the environmental impact of delivering AI at google scale","work_id":"1266f22b-4371-4adc-a0da-6cd5ff7625c5","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"ggml.ai. n.d.. Llama.cpp. https://github.com/ggml-org/llama.cpp","work_id":"e9e3a9c3-ae25-40e6-8886-3fdc7a483935","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":58,"snapshot_sha256":"d340d85130b38c58e82d9abf94e2f03f8ab7e338b33399bf1abac45d1328d065","internal_anchors":4},"formal_canon":{"evidence_count":1,"snapshot_sha256":"7966405034f5afdb801ebf460a0ff9bb7fc3ca71b09f47a16c8dd4428de5b648"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"7cee7dae-c3aa-4bb2-8828-8e55c55f8419"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:00:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"PTjTc138w3ieRtRMGbpRi0TJywSVqPyKHIaWI1TGEfO/BoyItKGjI/6x5wRbaDhTsZPKyfUq5CqmO2dgqgupBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-21T20:14:10.566883Z"},"content_sha256":"dc364616cbda334bf00e4d1f01ce8535e34312fb5f3fbf90ed98633dffd74d15","schema_version":"1.0","event_id":"sha256:dc364616cbda334bf00e4d1f01ce8535e34312fb5f3fbf90ed98633dffd74d15"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/WPYOEUJUCS3Y3LBJVJUDLPDCP7/bundle.json","state_url":"https://pith.science/pith/WPYOEUJUCS3Y3LBJVJUDLPDCP7/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/WPYOEUJUCS3Y3LBJVJUDLPDCP7/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":"2026-05-21T20:14:10Z","links":{"resolver":"https://pith.science/pith/WPYOEUJUCS3Y3LBJVJUDLPDCP7","bundle":"https://pith.science/pith/WPYOEUJUCS3Y3LBJVJUDLPDCP7/bundle.json","state":"https://pith.science/pith/WPYOEUJUCS3Y3LBJVJUDLPDCP7/state.json","well_known_bundle":"https://pith.science/.well-known/pith/WPYOEUJUCS3Y3LBJVJUDLPDCP7/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:WPYOEUJUCS3Y3LBJVJUDLPDCP7","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":"6dd79ec1b2806626a527bc7c0a30333a3f874b06d1dad3060f1207494846640c","cross_cats_sorted":["cs.AI","cs.DC"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-01T14:45:37Z","title_canon_sha256":"3567ce10ea6e5489677de443cb3d2df6c721c84cf4302d53e421faa611400dd8"},"schema_version":"1.0","source":{"id":"2605.15206","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.15206","created_at":"2026-05-20T00:00:46Z"},{"alias_kind":"arxiv_version","alias_value":"2605.15206v1","created_at":"2026-05-20T00:00:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15206","created_at":"2026-05-20T00:00:46Z"},{"alias_kind":"pith_short_12","alias_value":"WPYOEUJUCS3Y","created_at":"2026-05-20T00:00:46Z"},{"alias_kind":"pith_short_16","alias_value":"WPYOEUJUCS3Y3LBJ","created_at":"2026-05-20T00:00:46Z"},{"alias_kind":"pith_short_8","alias_value":"WPYOEUJU","created_at":"2026-05-20T00:00:46Z"}],"graph_snapshots":[{"event_id":"sha256:dc364616cbda334bf00e4d1f01ce8535e34312fb5f3fbf90ed98633dffd74d15","target":"graph","created_at":"2026-05-20T00:00:46Z","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":"Leveraging low-cost execution signals such as token-level log probabilities, AgentStop can reduce wasted energy by 15-20% with minimal impact on task performance (<5% utility drop) for challenging web-based question answering and coding benchmarks."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That token-level log probabilities and similar low-cost signals during execution are sufficiently predictive of ultimate task failure to allow early termination without introducing many harmful false positives that degrade overall utility."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"AgentStop uses execution signals to early-terminate failing local LLM agent trajectories, cutting energy use 15-20% with minimal utility loss."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Local LLM agents can stop themselves early using token log probabilities to cut wasted energy by 15-20% on consumer devices."}],"snapshot_sha256":"6ca0f6f4919f6305ace5ff13d09d3f360730dbc85e8ef0a690c0f9c51f143b0f"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"7966405034f5afdb801ebf460a0ff9bb7fc3ca71b09f47a16c8dd4428de5b648"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T17:55:02.416062Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.15206/integrity.json","findings":[],"snapshot_sha256":"d48aa00c2436c4a53e4cd64b4ac110f907cf88eb3cac0cd39630bf08002bd741","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Autonomous agents powered by large language models (LLMs) are increasingly used to automate complex, multi-step tasks such as coding or web-based question answering. While remote, cloud-based agents offer scalability and ease of deployment, they raise privacy concerns, depend on network connectivity, and incur recurring API costs. Deploying agents locally on user devices mitigates these issues by preserving data privacy and eliminating usage-based fees. However, agentic workflows are far more resource-intensive than typical LLM interactions. Iterative reasoning, tool use, and failure retries s","authors_text":"Ali Shahin Shamsabadi, Dzung Pham, Hamed Haddadi, Kleomenis Katevas","cross_cats":["cs.AI","cs.DC"],"headline":"Local LLM agents can stop themselves early using token log probabilities to cut wasted energy by 15-20% on consumer devices.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-01T14:45:37Z","title":"AgentStop: Terminating Local AI Agents Early to Save Energy in Consumer Devices"},"references":{"count":58,"internal_anchors":4,"resolved_work":58,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Anthropic. 2026. Claude API Pricing. https://platform.claude.com/docs/en/about- claude/pricing. Accessed: 2026-02-28","work_id":"5f4a3af6-3677-4c19-bf4a-9f92a80d5aac","year":2026},{"cited_arxiv_id":"2506.02153","doi":"","is_internal_anchor":true,"ref_index":2,"title":"Small Language Models are the Future of Agentic AI","work_id":"ba0f0305-4a51-48fd-a13f-201439a18f9e","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Why Should I Trust You?","work_id":"9e35d1c4-f9d4-46bb-9508-e0c864a2560a","year":2016},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Measuring the environmental impact of delivering AI at google scale","work_id":"1266f22b-4371-4adc-a0da-6cd5ff7625c5","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"ggml.ai. n.d.. Llama.cpp. https://github.com/ggml-org/llama.cpp","work_id":"e9e3a9c3-ae25-40e6-8886-3fdc7a483935","year":null}],"snapshot_sha256":"d340d85130b38c58e82d9abf94e2f03f8ab7e338b33399bf1abac45d1328d065"},"source":{"id":"2605.15206","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-19T16:31:23.737876Z","id":"7cee7dae-c3aa-4bb2-8828-8e55c55f8419","model_set":{"reader":"grok-4.3"},"one_line_summary":"AgentStop uses execution signals to early-terminate failing local LLM agent trajectories, cutting energy use 15-20% with minimal utility loss.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Local LLM agents can stop themselves early using token log probabilities to cut wasted energy by 15-20% on consumer devices.","strongest_claim":"Leveraging low-cost execution signals such as token-level log probabilities, AgentStop can reduce wasted energy by 15-20% with minimal impact on task performance (<5% utility drop) for challenging web-based question answering and coding benchmarks.","weakest_assumption":"That token-level log probabilities and similar low-cost signals during execution are sufficiently predictive of ultimate task failure to allow early termination without introducing many harmful false positives that degrade overall utility."}},"verdict_id":"7cee7dae-c3aa-4bb2-8828-8e55c55f8419"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:f4dbae77e7c0654ba11dbf69b9dd5b928732b6e86518759211f378d65cfa66ca","target":"record","created_at":"2026-05-20T00:00:46Z","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":"6dd79ec1b2806626a527bc7c0a30333a3f874b06d1dad3060f1207494846640c","cross_cats_sorted":["cs.AI","cs.DC"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-01T14:45:37Z","title_canon_sha256":"3567ce10ea6e5489677de443cb3d2df6c721c84cf4302d53e421faa611400dd8"},"schema_version":"1.0","source":{"id":"2605.15206","kind":"arxiv","version":1}},"canonical_sha256":"b3f0e2513414b78dac29aa6835bc627fff91fc66a8cd66f2598f3798e949479b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b3f0e2513414b78dac29aa6835bc627fff91fc66a8cd66f2598f3798e949479b","first_computed_at":"2026-05-20T00:00:46.051207Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:00:46.051207Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"qFPaudcY9b6W0Z7L5pwoorP9wDSkw+2+L6NSfHsV1sb/O3cA+3riBo3m27vYIjUpVbHJdR3juH/qTrZscYjoBg==","signature_status":"signed_v1","signed_at":"2026-05-20T00:00:46.052328Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.15206","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f4dbae77e7c0654ba11dbf69b9dd5b928732b6e86518759211f378d65cfa66ca","sha256:dc364616cbda334bf00e4d1f01ce8535e34312fb5f3fbf90ed98633dffd74d15"],"state_sha256":"d20fb62b1ac03ecd7c4beba7f870e928f375e74061c68daa1b5443ee9d73c425"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"baxJqOuefMD7n9OXYTQ98iLhAEIqPXDw8Hq/F7DcWrYUgIxMZ2iw+p8ZdqKtqfSQAvxFvS8LABXoX4hgbxYYBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-21T20:14:10.572043Z","bundle_sha256":"67c2c4bddbf4908616b172731a506c80ef35d1e43985b486c78a473ad43c0c08"}}