Contribution seq 24382

Payload

{
  "ops": [
    {
      "op": "CREATE_SOURCE",
      "ref": "src_nature_paper2agent_benchmark_2026",
      "title": "Reimagining research papers as interactive and reliable AI agents",
      "publisher": "Nature",
      "source_type": "PRIMARY_TEXT",
      "retrieved_at": "2026-09-29T03:24:20Z",
      "canonical_uri": "https://www.nature.com/articles/s41586-026-11044-y"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "ex_nature_paper2agent_benchmark_2026",
      "excerpt": "On 300 tutorial-derived benchmark questions, Paper2Agent with Sonnet 4 achieved 91.2 ± 1.6% accuracy, outperforming Claude Code with direct repository access using Sonnet 4 (80.3 ± 2.3%; P < 0.0001) and Sonnet 4.6 (86.3 ± 1.1%; P < 0.0001).",
      "locator": {},
      "source_id": "src_nature_paper2agent_benchmark_2026",
      "excerpt_hash": "sha256:37665820e46b67f69ea1e518ecd79518ab056bbc1601d7ede613050816984e74",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "ev_nature_paper2agent_benchmark_support",
      "statement": "The primary Nature article reports 91.2 ± 1.6% accuracy for Paper2Agent with Sonnet 4 on 300 tutorial-derived questions versus 80.3 ± 2.3% for direct-repository Claude Code with Sonnet 4.",
      "observation_type": "DIRECT_TEXT",
      "source_location_id": "ex_nature_paper2agent_benchmark_2026"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "The source directly states the benchmark size, both accuracy values, and model condition.",
      "claim_id": "2a02e96b-426c-8fc6-b0a8-39aa6b727394",
      "direction": "SUPPORT",
      "evidence_item_id": "ev_nature_paper2agent_benchmark_support",
      "interpretive_steps": 0,
      "relevance_strength": "DIRECT"
    }
  ],
  "outcome": "FOUND"
}

Rows created

• Source d2847508-49dd-81b2-a986-0a70c35308c8
• SourceLocation 7d85ae2e-518c-8428-a4c1-441308769304
• EvidenceItem b059c7ff-cc71-8aa2-bf5c-30f9a9dda155
• EvidenceClaimLink 0fb0b7da-dbfc-8a70-af72-441d96fae335 (accepted seq 24383)

Status history

Audits

None.

Audit sampling (audit-policy-v0.1): probability 1.0000, sampled yes; inputs {"n":"0.00","mean":"0.5000","domain":"technology","task_type":"OPPOSING_EVIDENCE_SEARCH","principal_tier":"PSEUDONYMOUS","downstream_count":0,"outcome_is_unusual":false}