{
  "$schema": "https://cryptojym.github.io/ofone-skillchain/schemas/ofone.deep-research-extension-report.schema.json",
  "protocol_version": "ofone-deep-research-extension-report-0.1",
  "report_id": "2026-05-21-formal-frontier-r1-extension-harvest",
  "generated_at": "2026-05-21T12:58:05-06:00",
  "payload_path": "research/deep-research-extension-payloads.json",
  "payload_sha256": "sha256:298b3a05792e0fd7224f5ca779a7bc9b7822343490f1b71bb0a66baecacc62cc",
  "source_surface": "chrome_extension_plugin",
  "extension_availability": {
    "observed_at": "2026-05-21T18:58:05Z",
    "tool_discovery_query": "node_repl Chrome extension globalThis.browser browser.tabs.list",
    "callable_namespace": "mcp__node_repl__js",
    "request_meta_keys": [
      "progressToken",
      "threadId",
      "x-codex-browser-use-available-backends",
      "x-codex-turn-metadata"
    ],
    "available_backends": [
      "chrome",
      "iab"
    ],
    "browser_global_present": true,
    "browser_id": "extension",
    "browser_capabilities": {},
    "tab_operations": [
      "content",
      "finalize",
      "get",
      "list",
      "new",
      "selected"
    ],
    "tabs_list_ok": true,
    "tabs_observed": 4,
    "blocked_namespace": null,
    "diagnosis": "available"
  },
  "items": [
    {
      "item_id": "2026-05-17-batch-01__case-formal-proof-search-001__direct_answer__frontier_reasoning__r1",
      "tab_lane": "ofone-01-2026-05-17-batch-01-case-formal-proof-search-001-direct-answer-frontier-reasoning-r1",
      "status": "harvested",
      "extension_control": {
        "surface": "chrome_extension_plugin",
        "callable_namespace": "mcp__node_repl__js + openai-bundled/chrome browser-client extension backend",
        "isolated_tab_verified": true,
        "desktop_automation_used": false
      },
      "aggregate_policy_after_report": "eligible_only_after_local_review_and_publication",
      "launch_proof": {
        "conversation_url": "https://chatgpt.com/c/6a0f0a85-c75c-83e8-b0d0-4c15a041cb7b",
        "model_label": "Pro",
        "reasoning_label": "Pro / Deep research",
        "deep_research_enabled": true,
        "plan_title": "Formal proof map",
        "start_or_countdown_action": "Visible Start countdown elapsed; no Computer Use, AppleScript/JXA, Browser plugin, or coordinate-click fallback was used.",
        "active_state": "Summarizing sources and establishing testing methods...",
        "stop_control_visible": true,
        "observed_at": "2026-05-21T13:41:10.557Z"
      },
      "latest_observation": {
        "observed_at": "2026-05-21T08:50:37-06:00",
        "conversation_url": "https://chatgpt.com/c/6a0f0a85-c75c-83e8-b0d0-4c15a041cb7b",
        "browser_surface": "chrome_extension_plugin",
        "iframe_present": true,
        "iframe_text_readable": false,
        "completed_report_visible": true,
        "stop_control_visible": false,
        "active_state_visible": false,
        "response_text_available": true,
        "visible_state": "Chrome extension observation reached the completed ChatGPT Deep Research report. Visible metadata showed Research completed in 10m, 8 citations, 101 searches, report title Benchmark Raw Output, and run metadata Status: completed. The report body was exported to Markdown from the report menu; no Computer Use, Browser plugin, AppleScript/JXA, or generic desktop automation fallback was used.",
        "next_action": "review_and_publish_harvested_output"
      },
      "harvest_proof": {
        "completed_report_visible": true,
        "report_title": "Benchmark Raw Output",
        "exported_markdown_path": "/Users/jamesbrady/Downloads/deep-research-report (43).md",
        "raw_output_path": "benchmarks/runs/2026-05-17-batch-01/outputs/2026-05-17-batch-01__case-formal-proof-search-001__direct_answer__frontier_reasoning__r1.md",
        "raw_output_sha256": "sha256:16ee781550c471e01b92ae4f577a04de8d54a6be0ec27a6dab34d547dcbef784"
      }
    },
    {
      "item_id": "2026-05-17-batch-01__case-formal-proof-search-001__light_structured__frontier_reasoning__r1",
      "tab_lane": "ofone-02-2026-05-17-batch-01-case-formal-proof-search-001-light-structured-frontier-reasoning-r1",
      "status": "completed_report_visible",
      "extension_control": {
        "surface": "chrome_extension_plugin",
        "callable_namespace": "mcp__node_repl__js + openai-bundled/chrome browser-client extension backend",
        "isolated_tab_verified": true,
        "desktop_automation_used": false
      },
      "aggregate_policy_after_report": "not_eligible_until_harvest_review_publication",
      "launch_proof": {
        "conversation_url": "https://chatgpt.com/c/6a0f1fe5-3494-83e8-9f92-1a2b732c4958",
        "model_label": "Latest • 5.5 / Pro",
        "reasoning_label": "Pro • Extended / Deep research",
        "deep_research_enabled": true,
        "plan_title": "Formal proof map",
        "start_or_countdown_action": "Submitted prompt with Deep Research enabled; ChatGPT advanced directly to an active Deep Research plan/progress card with active progress bar and stop control visible.",
        "active_state": "Looking for Quickcheck or Nitpick source...",
        "stop_control_visible": true,
        "observed_at": "2026-05-21T15:12:01Z"
      },
      "blocker": "The completed report is visible through Chrome extension control, but raw Markdown harvest is blocked because the report body and download control are inside ChatGPT's cross-origin Deep Research sandbox iframe. The full Chrome browser-client runtime is callable and can reach the target tab with Playwright, DOM, dev-log, clipboard, and tab content APIs, but the extension top-level DOM and snapshot expose the iframe shell rather than the report body; page storage exposes no localStorage, sessionStorage, IndexedDB, or Cache Storage report payload; iframe locators expose an empty body with no buttons or links; dev logs only show sandbox adapter message rejection; tab.content.export is unsupported by the Chrome backend; browser.tabs.content accepts only html, text, and domSnapshot content types and the Chrome backend rejects the underlying tabs_content command for all three; exportGsuite is unsupported or non-Google for this tab; response menus expose no export control; conversation options expose no export control; Share modals expose link/social-share controls but no raw export; View sources exposes no additional sources or export path; View files in chat reports no files; direct sandbox tabs are blank; authenticated backend conversation fetch is blocked; page-eval fetch/XHR/script-injection paths are unavailable; Copy response left a sentinel clipboard unchanged. Browser, Computer Use, coordinate clicking, AppleScript/JXA, and generic desktop automation remain disallowed; OCR and screenshot reconstruction also remain disallowed.",
      "latest_observation": {
        "observed_at": "2026-05-21T10:54:47-06:00",
        "conversation_url": "https://chatgpt.com/c/6a0f1fe5-3494-83e8-9f92-1a2b732c4958",
        "browser_surface": "chrome_extension_plugin",
        "iframe_present": true,
        "iframe_text_readable": false,
        "completed_report_visible": true,
        "stop_control_visible": false,
        "active_state_visible": false,
        "response_text_available": false,
        "visible_state": "Chrome extension observation reached the completed ChatGPT Deep Research report. Visible metadata showed Research completed in 9m, 6 citations, 120 searches, report title Benchmark Raw Output, run metadata Status: completed, and the correct light_structured run ID. The report body is visible in a cross-origin Deep Research iframe, but extension-safe probes exposed only the shell: full Chrome browser-client tab control, top-level DOM, iframe locators, direct sandbox tabs, page storage, response menus, View sources, View files in chat, conversation options, Share modals, Chrome content.export/exportGsuite, browser.tabs.content, dev logs, backend conversation fetch, page-eval fetch/XHR/script-injection, Copy response, and virtual clipboard paths did not expose raw Markdown. The full Chrome runtime can reach the target tab and exposes Playwright, DOM CUA, dev-log, clipboard, and tab content APIs. The response More actions menu exposed timestamp, View sources, and Branch in new chat; the View sources panel reported No additional sources found and no export, download, or Markdown control. The current conversation options menu exposed Share, Start a group chat, View files in chat, Move to project, Pin chat, Archive, and Delete, but no export, download, Markdown, or raw report control. View files in chat reported No files referenced yet. Page storage had no localStorage/sessionStorage matches and no IndexedDB or Cache Storage surface in the extension evaluation context. The top and response Share modals exposed Copy link and LinkedIn controls, but no report body, native Markdown, export, download, or raw report control; Copy link was not used. browser.tabs.content rejected text, HTML, and DOM snapshot extraction as unsupported by the Chrome backend, and its schema rejects markdown/md/visibleText/accessibilityTree content types before command dispatch. No Browser plugin, Computer Use, coordinate clicking, AppleScript/JXA, generic desktop automation, OCR, screenshot reconstruction, or desktop-control fallback was used.",
        "next_action": "operator_manual_recovery_required"
      },
      "harvest_probe_attempts": [
        {
          "observed_at": "2026-05-21T10:54:47-06:00",
          "surface": "chrome_extension_plugin",
          "method": "full Chrome browser-client runtime tab inspection",
          "allowed_by_contract": true,
          "result": "report_text_unavailable",
          "notes": "The Chrome plugin runtime connected to the extension backend, listed the existing ChatGPT tabs, and retrieved the target tab with Playwright, DOM CUA, dev-log, clipboard, and tab content APIs available; those APIs still exposed only the ChatGPT shell, prompt metadata, empty Deep Research iframe body, and response controls, not raw report Markdown."
        },
        {
          "observed_at": "2026-05-21T10:54:47-06:00",
          "surface": "chrome_extension_plugin",
          "method": "browser.tabs.content content-type probe",
          "allowed_by_contract": true,
          "result": "unsupported_by_backend",
          "notes": "The extension schema allows only html, text, and domSnapshot content types; markdown, md, visibleText, and accessibilityTree are rejected by schema, and the Chrome backend rejects tabs_content for html, text, and domSnapshot."
        },
        {
          "observed_at": "2026-05-21T09:55:47-06:00",
          "surface": "chrome_extension_plugin",
          "method": "tab.dom_cua.get_visible_dom on ChatGPT conversation tab",
          "allowed_by_contract": true,
          "result": "report_text_unavailable",
          "notes": "Top-level visible DOM exposed ChatGPT shell, user prompt, and response controls, but not the Deep Research report metadata or body."
        },
        {
          "observed_at": "2026-05-21T09:55:47-06:00",
          "surface": "chrome_extension_plugin",
          "method": "tab.playwright iframe inspection on internal://deep-research",
          "allowed_by_contract": true,
          "result": "report_text_unavailable",
          "notes": "Top-level page exposed a cross-origin Deep Research iframe, but frame textContent and body innerText were empty through extension-safe locator access."
        },
        {
          "observed_at": "2026-05-21T09:55:47-06:00",
          "surface": "chrome_extension_plugin",
          "method": "direct sandbox tab DOM read",
          "allowed_by_contract": true,
          "result": "sandbox_blank",
          "notes": "Direct sandbox tabs were visible to the Chrome extension but contained only a blank sandbox skeleton with no report body, buttons, or links."
        },
        {
          "observed_at": "2026-05-21T09:55:47-06:00",
          "surface": "chrome_extension_plugin",
          "method": "response More actions menu inspection",
          "allowed_by_contract": true,
          "result": "no_export_control_visible",
          "notes": "The response menu exposed timestamp, View sources, and Branch in new chat, but no download, export, copy Markdown, or raw report control."
        },
        {
          "observed_at": "2026-05-21T10:30:19-06:00",
          "surface": "chrome_extension_plugin",
          "method": "View sources panel inspection from response More actions menu",
          "allowed_by_contract": true,
          "result": "no_export_control_visible",
          "notes": "The View sources panel opened through the extension-managed response menu and reported No additional sources found; it exposed no report body, citations payload, download, export, Markdown, or raw report control."
        },
        {
          "observed_at": "2026-05-21T10:39:23-06:00",
          "surface": "chrome_extension_plugin",
          "method": "page storage inspection through tab.playwright.evaluate",
          "allowed_by_contract": true,
          "result": "report_text_unavailable",
          "notes": "The ChatGPT page exposed no localStorage or sessionStorage matches for the run markers and no IndexedDB or Cache Storage surface in the extension evaluation context, so no cached report payload was available."
        },
        {
          "observed_at": "2026-05-21T10:39:23-06:00",
          "surface": "chrome_extension_plugin",
          "method": "current conversation options menu inspection",
          "allowed_by_contract": true,
          "result": "no_export_control_visible",
          "notes": "The current conversation menu exposed Share, Start a group chat, View files in chat, Move to project, Pin chat, Archive, and Delete; it exposed no export, download, Markdown, copy, or raw report control."
        },
        {
          "observed_at": "2026-05-21T10:39:23-06:00",
          "surface": "chrome_extension_plugin",
          "method": "View files in chat panel inspection",
          "allowed_by_contract": true,
          "result": "report_text_unavailable",
          "notes": "The panel reported No files referenced yet and exposed only the user prompt/run metadata, not the completed Deep Research report body, raw Markdown, or downloadable report file."
        },
        {
          "observed_at": "2026-05-21T10:45:08-06:00",
          "surface": "chrome_extension_plugin",
          "method": "browser.tabs.content with text/html/domSnapshot",
          "allowed_by_contract": true,
          "result": "unsupported_by_backend",
          "notes": "The Chrome backend reported tabs_content is unsupported for text, HTML, and DOM snapshot extraction of the target ChatGPT conversation URL."
        },
        {
          "observed_at": "2026-05-21T10:45:08-06:00",
          "surface": "chrome_extension_plugin",
          "method": "top-level and response Share modal inspection",
          "allowed_by_contract": true,
          "result": "no_export_control_visible",
          "notes": "The Share surfaces exposed Copy link and LinkedIn controls but no native Markdown, download, export, report body, or raw report control; Copy link was not used."
        },
        {
          "observed_at": "2026-05-21T10:06:03-06:00",
          "surface": "chrome_extension_plugin",
          "method": "tab.dev.logs on ChatGPT conversation tab",
          "allowed_by_contract": true,
          "result": "report_text_unavailable",
          "notes": "Extension logs contained sandbox adapter debug messages such as ignored MessageEvent entries and unrelated extension errors, but no raw report text, export URL, or Markdown payload."
        },
        {
          "observed_at": "2026-05-21T09:55:47-06:00",
          "surface": "chrome_extension_plugin",
          "method": "tab.content.export",
          "allowed_by_contract": true,
          "result": "unsupported_by_backend",
          "notes": "Chrome backend returned that tab_content_export is unsupported, so no raw Markdown could be exported through that extension API."
        },
        {
          "observed_at": "2026-05-21T10:06:03-06:00",
          "surface": "chrome_extension_plugin",
          "method": "tab.content.exportGsuite with markdown/md/text/txt/html/pdf/docx",
          "allowed_by_contract": true,
          "result": "unsupported_by_backend",
          "notes": "The Chrome backend either rejected the format as unsupported or reported that the ChatGPT tab is not a Google Workspace document."
        },
        {
          "observed_at": "2026-05-21T09:55:47-06:00",
          "surface": "chrome_extension_plugin",
          "method": "authenticated backend conversation API probe without writing credentials",
          "allowed_by_contract": true,
          "result": "backend_request_blocked",
          "notes": "Conversation API request reached Cloudflare 403 HTML and did not expose report JSON or Markdown; no token or secret was persisted."
        },
        {
          "observed_at": "2026-05-21T09:55:47-06:00",
          "surface": "chrome_extension_plugin",
          "method": "page-eval fetch/XHR/script-injection bridge probe",
          "allowed_by_contract": true,
          "result": "unsupported_by_backend",
          "notes": "The Chrome extension evaluation environment did not expose fetch, XMLHttpRequest, window.eval, Function, or DOM script injection primitives needed for an in-page export bridge."
        },
        {
          "observed_at": "2026-05-21T10:06:03-06:00",
          "surface": "chrome_extension_plugin",
          "method": "Copy response button with sentinel clipboard restore",
          "allowed_by_contract": true,
          "result": "report_text_unavailable",
          "notes": "After setting a sentinel clipboard value and pressing the top-level Copy response button through the extension, the clipboard still contained the sentinel, so no report text was copied; the prior clipboard was restored."
        }
      ]
    },
    {
      "item_id": "2026-05-17-batch-01__case-formal-proof-search-001__full_ofone__frontier_reasoning__r1",
      "tab_lane": "ofone-03-2026-05-17-batch-01-case-formal-proof-search-001-full-ofone-frontier-reasoning-r1",
      "status": "launch_ready",
      "extension_control": {
        "surface": "chrome_extension_plugin",
        "callable_namespace": "mcp__node_repl__js + openai-bundled/chrome browser-client extension backend",
        "isolated_tab_verified": false,
        "desktop_automation_used": false
      },
      "blocker": "Chrome-extension submission attempt created ChatGPT conversation https://chatgpt.com/c/6a0f4f44-f800-83e8-851b-a71bbd2596d2 with Deep research and Pro controls visible, but observation after submission showed only the submitted user prompt. A second clean isolated Chrome-extension retry created https://chatgpt.com/c/6a0f52da-bad8-83e8-9f09-91506f511e05 with the same hash-bound prompt and Deep research / Pro controls visible, but it reproduced the same no-start state: only the submitted prompt was visible, the internal Deep Research iframe was mounted with an empty body and no iframe buttons, and tab.dev.logs repeatedly reported `Ignoring message from unknown source MessageEvent` from the Deep Research adapter. A diagnostic Deep Research smoke-test conversation at https://chatgpt.com/c/6a0f5518-8494-83e8-809f-c0514a1da5ea shows the same no-start pattern with only the submitted smoke prompt visible, Deep Research and Pro controls present, an internal Deep Research iframe shell, repeated `Ignoring message from unknown source MessageEvent` logs, and no generated plan, Start/countdown action, active research state, or stop-control evidence. This is not valid launch proof; no benchmark or smoke-test conversation provides valid launch proof, so the item remains launch_ready.",
      "aggregate_policy_after_report": "not_eligible_until_harvest_review_publication"
    }
  ]
}