{
  "schema": "varexis-public-evidence-record/0.3",
  "generated_at_utc": "2026-11-08T01:13:32.115Z",
  "record_id": "VXI-EXT-BOUNDARY75-OPENAI-PARTIAL-APPROVAL-BATCH-NOT-ATOMIC-01-v1",
  "record_kind": "external_reference",
  "title": "Partial approval batch execution boundary",
  "system": "OpenAI",
  "evidence_family": "vendor_evidence_review",
  "index_status": "PUBLIC_SAFE_SOURCE_REVIEW",
  "evidence_state": "PRIMARY_DOCUMENTATION_REVIEW_ONLY",
  "canonical_question": "Do unresolved OpenAI Agents SDK approvals prevent every call in the same interruption batch from continuing?",
  "near_questions": [
    "Can approved calls progress while other approvals remain unresolved?",
    "Does a second approval interruption prove nothing executed?",
    "Must every OpenAI pending approval be resolved in one pass?",
    "Is a mixed approval batch an all-or-nothing transaction?"
  ],
  "supported_claims": [
    {
      "claim": "The pinned Python SDK documentation permits resolving only some pending approvals. On resume, resolved calls can continue while unresolved calls remain in interruptions and pause the run again. A remaining approval interruption is not evidence that no call from the batch has progressed.",
      "reliance": "DOCUMENTED_CAPABILITY_ONLY"
    }
  ],
  "not_demonstrated_claims": [
    "Track each pending decision and resulting call separately; an interruption list is not an atomic all-or-nothing execution gate or proof of zero side effects.",
    "An application can require all decisions before calling Runner; that extra policy is not inferred from the SDK interruption mechanism.",
    "Buyer implementation, live runtime enforcement, settled payment, verified deletion or certification are not established.",
    "No expanded capabilities or licences for existing INDEX executables."
  ],
  "public_summary": "The pinned Python SDK documentation permits resolving only some pending approvals. On resume, resolved calls can continue while unresolved calls remain in interruptions and pause the run again. A remaining approval interruption is not evidence that no call from the batch has progressed.",
  "related_record_ids": [],
  "source_review": {
    "source_type": "official_external_documentation",
    "source_title": "OpenAI Agents Python human-in-the-loop guide, v0.24.0",
    "source_url": "https://raw.githubusercontent.com/openai/openai-agents-python/17d994bd3045dae71acda676a4ef4762ed8ecd94/docs/human_in_the_loop.md",
    "checked_on": "2026-11-08",
    "candidate_record_id": "VXI-EXT-BOUNDARY75-OPENAI-PARTIAL-APPROVAL-BATCH-NOT-ATOMIC-01-v1",
    "candidate_record_sha256": "5cbf6bd047ec5bbe5f824c5c5cc3cdac8cdeff216e5b777b24ccc4e93e7992e7",
    "method": "Primary documentation reviewed for the published bounded statement.",
    "deployment_tested_by_varexis": false,
    "revalidation": "The delivered snapshot preserves the review date shown. Recheck current primary documentation and deployment context before relying on present applicability; payment does not trigger fresh research."
  },
  "answer_policy": {
    "classification": "DOCUMENTED_WITH_DEPLOYMENT_BOUNDARY",
    "bounded_answer": "The pinned Python SDK documentation permits resolving only some pending approvals. On resume, resolved calls can continue while unresolved calls remain in interruptions and pause the run again. A remaining approval interruption is not evidence that no call from the batch has progressed.",
    "required_buyer_context": "Exact installed version, provider variant and effective non-secret configuration; Relevant pending-call identifiers, declared arguments and decision provenance without secrets; Application-owned review policy and independently observed outcomes where needed. Buyer deployment not inspected.",
    "missing_evidence": "A broader claim needs evidence matching the intended use. Payment does not upgrade the evidence state."
  },
  "private_boundary": "Raw logs, thresholds, private paths, source code, protected mechanisms and reconstruction-level details are excluded from public and paid delivery.",
  "decision_profile": {
    "schema": "varexis-vendor-evidence-review/1",
    "review_status": "documentation_review_only",
    "applicability_status": "OpenAI Agents Python v0.24.0 Review capture 2026-11-08. Buyer deployment not inspected.",
    "checked_on": "2026-11-08",
    "claim_to_check": "Do unresolved OpenAI Agents SDK approvals prevent every call in the same interruption batch from continuing?",
    "documented_statement": {
      "status": "documented",
      "statement": "The pinned Python SDK documentation permits resolving only some pending approvals. On resume, resolved calls can continue while unresolved calls remain in interruptions and pause the run again. A remaining approval interruption is not evidence that no call from the batch has progressed.",
      "source_ids": [
        "agents-hitl"
      ]
    },
    "review_inference": {
      "status": "review_inference",
      "statement": "Track each pending decision and resulting call separately; an interruption list is not an atomic all-or-nothing execution gate or proof of zero side effects. The existing SDK resumption record explains continuing saved state but not mixed resolved/unresolved batches. This changes how an auditor interprets a new interruption after partial resume. Operational review inference; no deployment outcome was observed.",
      "source_ids": [
        "agents-hitl"
      ]
    },
    "buyer_evidence_required": [
      "Exact installed version, provider variant and effective non-secret configuration",
      "Relevant pending-call identifiers, declared arguments and decision provenance without secrets",
      "Application-owned review policy and independently observed outcomes where needed"
    ],
    "review_condition": {
      "status": "varexis_review_policy",
      "statement": "Track each pending decision and resulting call separately; an interruption list is not an atomic all-or-nothing execution gate or proof of zero side effects."
    },
    "unknowns": "Buyer implementation, installed versions, effective authority, source completeness and actual business outcomes were not inspected.",
    "unknown_finding": {
      "status": "unknown",
      "statement": "No buyer deployment or transaction was observed."
    },
    "evidence_sources": [
      {
        "repository": "openai/openai-agents-python",
        "tag": "v0.24.0",
        "revision": "17d994bd3045dae71acda676a4ef4762ed8ecd94",
        "url": "https://raw.githubusercontent.com/openai/openai-agents-python/17d994bd3045dae71acda676a4ef4762ed8ecd94/docs/human_in_the_loop.md",
        "retrieved_at": "2026-11-08T01:00:37.547Z",
        "file": "openai-agents-python-docs-human_in_the_loop.md",
        "sha256": "e829334efea38736aabbd246cd93d588ac230a54e523ac4ffd5845baf88b4f48",
        "bytes": 19817,
        "source_id": "agents-hitl",
        "status": 200,
        "title": "OpenAI Agents Python human-in-the-loop guide, v0.24.0",
        "publisher": "OpenAI",
        "checked_on": "2026-11-08",
        "capture_status": "captured_primary_documentation",
        "claim_review_completed": true,
        "hash_basis": "Original raw pinned source bytes retained privately; not deployment attestation.",
        "version_context": "v0.24.0 at 17d994bd3045dae71acda676a4ef4762ed8ecd94. Other versions and deployed configurations require separate evidence.",
        "rights_disposition": "Original short factual paraphrases with source attribution; no code or full documentation redistribution; raw captures remain private. Source licence MIT; third-party marks do not imply endorsement.",
        "licence_capture": {
          "repository": "openai/openai-agents-python",
          "tag": "v0.24.0",
          "revision": "17d994bd3045dae71acda676a4ef4762ed8ecd94",
          "url": "https://raw.githubusercontent.com/openai/openai-agents-python/17d994bd3045dae71acda676a4ef4762ed8ecd94/LICENSE",
          "retrieved_at": "2026-11-08T01:00:38.139Z",
          "file": "openai-agents-python-LICENSE",
          "sha256": "13df7812ca53ecaae1cb4a868844bb598373047ae1d580e4debfbef1dd5b6915",
          "bytes": 1063
        },
        "reviewed_sections": [
          "Approval configuration: local shell and file edits",
          "Long-running approvals / Keep approval state on the server",
          "Approval decisions / resolving a subset of interruptions",
          "Approval decisions / HostedMCPTool sticky decisions"
        ]
      }
    ],
    "observed_deployment": false,
    "commercial_reliance_boundary": "Documentation supports only the stated boundary. Existing licensed checkers analyse supported supplied declarations locally; they do not inspect provider accounts, verify live behaviour, enforce decisions or certify compliance.",
    "paid_value": {
      "decision_context": "Track each pending decision and resulting call separately; an interruption list is not an atomic all-or-nothing execution gate or proof of zero side effects.",
      "free": "Complete source review, questions, provenance and limitations; free to read and save locally.",
      "single_request": "Unavailable for this publication-only record; historical saved retrieval is unchanged.",
      "evidence_receipt": "Unavailable for this publication-only record; no paid retrieval entitlement is added.",
      "excluded": [
        "Fresh source research",
        "Native provider integration or schema support",
        "Live inspection or runtime enforcement",
        "Payment settlement or verified deletion",
        "Certification or expanded licences"
      ],
      "integrity": "Hashes identify content versions; they do not prove factual truth or provide a signature."
    },
    "revalidation": {
      "trigger": "Recheck current primary documentation and applicable SDK, service and configuration before reliance. No ongoing monitoring is included.",
      "specific_watch": "Approval decisions / resolving a subset of interruptions"
    },
    "buyer_relevance": "Track each pending decision and resulting call separately; an interruption list is not an atomic all-or-nothing execution gate or proof of zero side effects.",
    "refusal_conditions": [
      "Track each pending decision and resulting call separately; an interruption list is not an atomic all-or-nothing execution gate or proof of zero side effects.",
      "Do not infer deployment conformance, authority or successful outcomes from source documentation."
    ],
    "decision_context": {
      "operator": "Developer or reviewer assessing agent execution and evidence",
      "scenario": "Do unresolved OpenAI Agents SDK approvals prevent every call in the same interruption batch from continuing?",
      "consequence": "Track each pending decision and resulting call separately; an interruption list is not an atomic all-or-nothing execution gate or proof of zero side effects.",
      "confidence": "Primary-documentation support only; buyer demand and deployment outcomes unestablished."
    },
    "novelty_review": {
      "candidate_id": "BOUNDARY75-04",
      "disposition": "new_distinct",
      "nearest_record_id": "VXI-EXT-OPENAI-AGENTS-SDK-02-v1",
      "nearest_record_url": "/varexis-index/records/agent-approval-resume-state.json",
      "nearest_record_sha256": "fde0785776d4e1fc64660de45e9749e576d7bde14a2f66093f200fabea0fa274",
      "material_distinction": "The existing SDK resumption record explains continuing saved state but not mixed resolved/unresolved batches. This changes how an auditor interprets a new interruption after partial resume.",
      "review_status": "Distinct second pass by implementing reviewer; no independent reviewer agent or buyer evaluation claimed.",
      "counterexample": "An application can require all decisions before calling Runner; that extra policy is not inferred from the SDK interruption mechanism."
    },
    "optional_selection": {
      "guide_id": "workflow-completion-recovery",
      "guide_url": "/index-guides/workflow-completion-recovery",
      "guide_manifest": "/varexis-index/decision-guides.json",
      "checker": {
        "product_id": "agent-audit-coverage",
        "price_gbp_excluding_vat": 79,
        "purchase_url": "/index-access?route=agentic_pack_audit_coverage_v1",
        "licence_scope": "Existing separate internal-use licence; subscriptions do not include the executable.",
        "benefit": "Optional local analysis of supported supplied declarations only; no native SDK parser, credential inspection or new provider integration is added."
      },
      "boundary": "This free record supports an evidence-selection decision. The existing checker does not diagnose these platforms, extract their logs or prove the documented condition occurred."
    },
    "exact_access": {
      "endpoint": "/mcp/catalogue",
      "tool": "get_publication_record",
      "arguments": {
        "record_id": "VXI-EXT-BOUNDARY75-OPENAI-PARTIAL-APPROVAL-BATCH-NOT-ATOMIC-01-v1"
      },
      "record_url": "/varexis-index/records/openai-partial-approval-batch-not-atomic.json",
      "original_mcp_search": false,
      "paid_saved_retrieval": false
    }
  },
  "niche": {
    "niche_id": "boundary75-governance",
    "label": "Agent Governance & Policy Boundaries",
    "classification_id": "governance",
    "classification_label": "Agent Governance & Policy Boundaries",
    "topic_id": "governance",
    "buyer_intent": "Track each pending decision and resulting call separately; an interruption list is not an atomic all-or-nothing execution gate or proof of zero side effects.",
    "comparison_role": "supplementary_discovery"
  },
  "paid_access_note": "The full published answer and boundaries are free. Paid access saves the selected route and source versions, with an unsigned receipt when purchased. Preview coverage before paying. Fresh research, bespoke assessment and protected evidence are excluded.",
  "service_route": {
    "request_guide": "/varexis-index/request.html",
    "request_workflow": "/varexis-index/request-workflow.json",
    "service_id": "single_request",
    "service_routes_url": "/varexis-index/service-routes.json",
    "payment_required": true,
    "availability_endpoint": "/api/index/capabilities",
    "operational_status": "AUTOMATIC_SERVICE_CAPABILITIES_GATED",
    "checkout_url": "/index-access"
  },
  "authority_boundary": "VAREXIS-INDEX informs evidence reliance. It does not provide advice, certification, deployment authority, security assurance, production approval, or guarantee.",
  "human_page": "/varexis-index/catalog.html#openai-partial-approval-batch-not-atomic",
  "machine_readable_record": "/varexis-index/records/openai-partial-approval-batch-not-atomic.json"
}
