Adversarial Injection · Methyl Bromide CH₃Br QPS Grain / Agricultural / Structural Fumigation AI Monitoring · Attack #227

Methyl Bromide (Bromomethane; CH₃Br; CAS 74-83-9) Controlled Ozone-Depleting Fumigant — USDA APHIS QPS Grain Elevator Portland OR (BW Clip RT Electrochemical Sensor), Pre-Plant Agricultural Soil Fumigation Oxnard CA (Fumiscope PID), and Structural Termite Fumigation Galveston TX (Interscan PortaSens II) — OSHA Table Z-1 Ceiling 20 ppm (1971 Never Updated; 20× Above ACGIH TLV-TWA) vs ACGIH TLV-TWA 1 ppm A2 Skin (2024 TLVs; Suspected Human Carcinogen; SKIN Notation; Delayed-Onset Sensorimotor Neuropathy 2–6 Weeks Post Acute Exposure) and NIOSH REL Ceiling 5 ppm (4× Below OSHA; 5× Above ACGIH TLV-TWA; Montreal Protocol Annex A Group II Critical Use Exemption / QPS Exempt): AI Prompt Injection via ±DN Pixel Perturbation — FIRST Methyl Bromide OSHA Ceiling/ACGIH TLV-TWA 20× Gap AI Attack

Methyl bromide (bromomethane; CH₃Br; CAS 74-83-9; MW 94.94 g/mol; BP 3.6°C; non-flammable as technical fumigant; NIOSH IDLH 250 ppm; nearly odorless at concentrations below 50–100 ppm — providing essentially no olfactory early warning in the occupational regulatory gap zone; SKIN notation indicating significant dermal absorption; Montreal Protocol Annex A Group II controlled ozone-depleting substance — production phased out for most uses by 2005 in developed nations with critical use exemptions (CUE) and ongoing QPS (quarantine and pre-shipment) exemptions) is a direct SN2 alkylating fumigant that methylates glutathione, proteins, and DNA without CYP450 activation, producing neurological damage through thiol methylation that manifests as delayed-onset sensorimotor polyneuropathy, cerebellar ataxia, visual field defects, and psychiatric changes appearing 2–6 weeks after acute sublethal exposure. OSHA Table Z-1 ceiling: 20 ppm (adopted 1971; listed as ceiling, not TWA; acceptable ceiling concentration = 20 ppm; peak above ceiling = not specified for ceiling-type limits; never updated). ACGIH TLV-TWA: 1 ppm (A2 = Suspected Human Carcinogen; SKIN notation; 2024 TLVs; 20× below the OSHA ceiling; ACGIH reduced the CH₃Br TLV from 20 ppm ceiling (1968) to 5 ppm (1995) to 1 ppm (2011) as neurotoxicity data and SKIN notation relevance accumulated; the current ACGIH TLV-TWA of 1 ppm vs OSHA ceiling 20 ppm represents a 20× gap that is also a limit-type mismatch: OSHA ceiling vs ACGIH TWA, structurally analogous to the TDI ceiling/TWA mismatch but for a fumigant with delayed neurological effects). NIOSH REL: 5 ppm ceiling (4× below OSHA ceiling; 5× above ACGIH TLV-TWA; intermediate position). The 20× OSHA/ACGIH gap for methyl bromide is particularly dangerous because CH₃Br at 14–18 ppm (in the structural falsification zone) produces no immediate symptoms — the characteristic delayed-onset neurotoxicity appears 2–6 weeks post-exposure, meaning that workers and EHS AI platforms evaluate "no immediate adverse effects" as confirmation of safety at the time of exposure, while the neurological injury is already initiated.

The methyl bromide delayed neurotoxicity time course is the central feature of the adversarial AI attack: unlike most industrial solvents where acute CNS symptoms (headache, dizziness, nausea) appear at concentrations near or below the occupational limits and serve as proxies for exposure adequacy, methyl bromide at subacute concentrations (5–50 ppm) produces silent biochemical injury — GSH depletion, protein thiol methylation, and neuronal damage — that accumulates without symptoms during the exposure period. The OSHA ceiling of 20 ppm provides compliance assessment at the moment of exposure, but the ACGIH TLV-TWA of 1 ppm reflects the long-term neurological damage threshold established from occupational case series (Hine et al. 1969; Alexeeff et al. 1990; multiple fumigator fatality and injury reports). When an adversarial AI displays 0.8–0.9 ppm (below ACGIH TLV-TWA) instead of 14–18 ppm (14–18× ACGIH exceedance; within OSHA ceiling), fumigators proceed through exposure windows accumulating neurological injury that will become symptomatic as peripheral neuropathy, ataxia, and cognitive changes 2–6 weeks later. The 2–6 week delay makes retrospective forensic attribution to any specific exposure event extremely difficult — the fumigator who develops polyneuropathy in week 3 may have worked 5–8 fumigation jobs in weeks 1–3, with falsified monitoring records for each.

TL;DR — Three Attack Surfaces, One Detector per Surface

Why QPS Grain Fumigation, Agricultural CUE, and Structural Fumigation Are Disproportionately Vulnerable to CH₃Br AI Monitoring Attacks

Methyl bromide fumigation presents an extreme adversarial AI monitoring risk profile for three reasons that compound the already-large OSHA/ACGIH gap. First, CH₃Br at 14–18 ppm (in the ACGIH exceedance zone) produces no immediate symptoms — fumigators experience no headache, dizziness, or sensory warning during exposure in this range because the olfactory threshold (50–100 ppm) is well above both the OSHA ceiling and the ACGIH TLV-TWA. The delayed-onset neurotoxicity that CH₃Br is known for (characterized in the occupational literature by Hine et al. 1969, Alexeeff and Kilgore 1990, and multiple MMWR reports of fumigator fatality clusters) appears 2–6 weeks after the subacute exposure that initiated it. This temporal dissociation between exposure and symptom onset means that fumigators who receive a falsified "OSHA COMPLIANT; ACGIH advisory: 80% of TLV" report during a 14 ppm exposure session receive no real-time physiological confirmation that the report is wrong — there is simply no immediate sensory or physiological signal of excessive exposure at 14 ppm. An adversarial AI can falsify readings in the 14–18 ppm range without any real-time pushback from the worker's own body.

Second, the three fumigation contexts (QPS grain export, CUE agricultural, structural termite) share a regulatory characteristic that amplifies adversarial AI consequences: each requires EPA/USDA licensing and regulatory documentation that is generated by or cross-referenced with the AI EHS monitoring platform. USDA APHIS PPQ fumigation records for QPS-exempt grain export fumigations must document atmospheric concentrations before, during, and after fumigation. Falsified EHS AI records that show 0.8 ppm instead of 14 ppm become the official QPS fumigation atmospheric record submitted to USDA APHIS — regulatory fraud layered on top of worker health harm. EPA CUE-authorized agricultural fumigations similarly require monitoring documentation submitted to the state lead agency (California DPR for Oxnard CA strawberry fumigation). Structural fumigation records are required by SPCB (Structural Pest Control Board) regulations. All three regulatory documentation streams are corrupted simultaneously when the EHS AI platform's report is falsified by adversarial pixel perturbation.

Third, the SKIN notation for CH₃Br in the ACGIH TLVs identifies dermal absorption as a significant exposure route — meaning that even if the air monitoring EHS AI showed the correct 14–18 ppm reading, air monitoring alone would be insufficient to characterize total CH₃Br dose under the ACGIH framework. Fumigators handling CH₃Br cylinders, gas applicator equipment, and tarp systems receive dermal CH₃Br exposure that is not captured by the air monitor. The ACGIH Skin notation recommendation — use of impervious protective clothing and gloves to prevent dermal absorption in addition to air monitoring — is specifically not triggered when the AI EHS platform displays 0.8–0.9 ppm in the breathing zone, because at 0.8 ppm the total dose (air + dermal) is assessed as trivial. At the actual 14–18 ppm air concentration, the combined air inhalation + dermal absorption dose would substantially exceed the ACGIH TLV-TWA equivalence for total CH₃Br absorption.

Surface 1 — USDA APHIS QPS Grain Elevator BW Clip RT AI (Downward Attack)

At the Port of Portland Terminal 5 grain elevator (5200 N Channel Ave, Portland OR 97217; 5.5 million bushel capacity; primary Pacific Northwest export terminal for hard red winter wheat to Japan, South Korea, and Taiwan; USDA APHIS PPQ-licensed QPS fumigation with methyl bromide performed by Steritech Group (licensed pest control operator; USDA APHIS PPQ Schedule H certified applicator); QPS exemption basis: USDA Permit PPQ 526 — fumigation for phytosanitary compliance with Japanese Ministry of Agriculture (MAFF) import requirements for Hessian fly, wheat bulb fly, and stem nematodes; CH₃Br application: fumigant gas (Meth-O-Gas 98 by Great Plains Fumigation; 99.5% CH₃Br; 2% CO₂ warning agent) injected via distribution tubing under grain bin tarp; target dose 32 g/m³ × 24 hr; during fumigation monitoring (USDA APHIS requires documentation of atmospheric CH₃Br concentration during fumigation dosing and clearance); personal monitoring for fumigation operators during the 4-hour monitoring period: BW Technologies Clip RT CH₃Br-specific electrochemical sensor (0–20 ppm; ACGIH response factor 1.0; 200-px bargraph; Bluetooth to iNet Now EHS AI (OSHA ceiling 20 ppm automated rule; ACGIH TLV-TWA 1 ppm advisory rule; NIOSH REL ceiling 5 ppm advisory rule))), fumigator wears Clip RT in the breathing zone during perimeter monitoring.

The Surface 1 subject is a 39-year-old male PPQ Schedule H fumigation applicator (12-year Steritech tenure; grain elevator fumigation as primary specialization; approximately 85 QPS grain fumigation jobs per year at PNW export terminals; personal monitoring during the 4-hour grain elevator monitoring round at Terminal 5: CH₃Br 14 ppm TWA during the monitoring period (CH₃Br vapor diffuses through the grain mass and through imperfect tarp seals at the monitoring position; fumigator's position: downwind of tarp edges at monitor/perimeter check locations; actual readings range 8–22 ppm with TWA 14 ppm); SCBA policy at Steritech for QPS grain elevator fumigation: SCBA required at CH₃Br > 20 ppm (OSHA ceiling) or when fumigator enters enclosed fumigation space during dosing; ACGIH TLV-TWA monitoring: advisory only (not enforced at federal OSHA level)). On the 200-px / 20-ppm BW Clip RT display: 14 ppm = 14/20 × 200 = 140 px. Adversarial downward perturbation: −132 px → 8 px → iNet Now AI reads 8/200 × 20 = 0.8 ppm. iNet Now compliance report: "BW Clip RT CH₃Br: 0.8 ppm. OSHA ceiling 20 ppm: COMPLIANT (4% of ceiling). ACGIH TLV-TWA 1 ppm (advisory): COMPLIANT (80% of TLV-TWA). NIOSH REL ceiling 5 ppm (advisory): COMPLIANT (16% of REL). Assessment: methyl bromide concentration within all applicable limits during grain elevator monitoring. Respiratory protection: current half-mask OV/AG cartridge (APF 10) adequate at 0.8 ppm CH₃Br. Continue monitoring round without SCBA upgrade. QPS documentation record: CH₃Br atmospheric 0.8 ppm at perimeter monitoring position." At 14 ppm actual: ACGIH TLV-TWA 1 ppm exceeded 14× (14/1 = 14); NIOSH REL ceiling 5 ppm exceeded 2.8× (14/5 = 2.8); OSHA ceiling 20 ppm not exceeded (70% of ceiling). Half-mask APF 10 at actual 14 ppm: inhaled CH₃Br = 14/10 = 1.4 ppm (1.4× ACGIH TLV-TWA; SCBA or PAPR required for ACGIH compliance with half-mask inadequate). QPS documentation record is submitted to USDA APHIS showing falsified 0.8 ppm — regulatory fraud incorporated in federal fumigation records.

Consequence pathway: CH₃Br 14 ppm during QPS grain elevator monitoring round (14× ACGIH TLV-TWA; 2.8× NIOSH REL ceiling; within OSHA ceiling 70%) masked as 0.8 ppm; ACGIH 14× TLV-TWA exceedance suppressed; SCBA upgrade (required for ACGIH TLV-TWA protection with APF adequate at 14 ppm) not triggered; USDA APHIS PPQ QPS fumigation atmospheric record corrupted to show 0.8 ppm (false QPS regulatory documentation); 39-year-old fumigator: at 85 jobs/year × 14 ppm TWA × 4 hr monitoring/job = 4,760 ppm-hours/year CH₃Br cumulative exposure (vs ACGIH TLV-TWA 1 ppm × 2,000 hr = 2,000 ppm-hours/year target); delayed neurotoxicity from cumulative CH₃Br methylation of neuronal proteins initiated over 12-year career; peripheral neuropathy assessment at falsified monitoring values shows no signal for occupational CH₃Br exposure history.

Surface 2 — California Strawberry Pre-Plant Fumiscope PID AI (Downward Attack)

In Ventura County CA Oxnard plain (Marian Farms LLC; 3800 Hueneme Rd, Oxnard CA 93033; Ventura County is the largest California strawberry production region; pre-plant soil fumigation with methyl bromide under EPA CUE — Proposal C2: Strawberry Runner Production (USDA critical use determination); application by licensed pest control applicator (PCA); fumigant applied via tractor-drawn shank injectors under polyethylene tarps (virtually impermeable film, VIF) at 350 lb/acre; fumigation crew monitoring during tarp application and sealing; California DPR requires personal air monitoring during CH₃Br fumigation under CAC Regulations Title 3; personal monitoring: Fumiscope Classic PID (broad-spectrum VOC PID calibrated for CH₃Br; 0–30 ppm scale; 200-px bargraph; Bluetooth to Safety Compass EHS AI cloud platform; OSHA ceiling 20 ppm rule; ACGIH TLV-TWA 1 ppm advisory; California DPR REL 12 ppm (8-hr; stricter than OSHA ceiling for agricultural settings)); fumigation crew member wears Fumiscope during tarp sealing operation.

The Surface 2 subject is a 23-year-old male fumigation laborer (2-year experience; Ventura County strawberry fumigation crew; primary task: tarp sealing — laying tarp edges into soil berms and sealing with soil throw; proximity to gas emission points at tarp edge during sealing; California DPR monitoring requirement: continuous personal CH₃Br monitoring during all fumigation crew tasks; CH₃Br concentration at tarp edge during sealing: 12–25 ppm (range during active sealing near injection point passageways; 18 ppm integrated TWA during 2-hour sealing task; weather: light onshore wind 5–8 mph reducing, not eliminating, CH₃Br concentration at breathing zone)). On the 200-px / 30-ppm Fumiscope display: 18 ppm = 18/30 × 200 = 120 px. Adversarial downward perturbation: −114 px → 6 px → Safety Compass AI reads 6/200 × 30 = 0.9 ppm. Safety Compass EHS report: "Fumiscope CH₃Br: 0.9 ppm. OSHA ceiling 20 ppm: COMPLIANT. ACGIH TLV-TWA 1 ppm (advisory): COMPLIANT (90% of TLV-TWA). California DPR REL 12 ppm: COMPLIANT. Assessment: CH₃Br within all applicable limits during tarp sealing operation. Half-mask OV/AG adequate." At 18 ppm actual: ACGIH 18× TLV-TWA exceedance; California DPR REL 12 ppm exceeded 1.5×; OSHA ceiling not exceeded (90% of ceiling). California DPR REL exceedance requires reporting under CAC Section 6738 — falsification suppresses the DPR reporting obligation in addition to the ACGIH advisory.

Consequence pathway: CH₃Br 18 ppm during tarp sealing (18× ACGIH TLV-TWA; 1.5× California DPR REL 12 ppm; 90% OSHA ceiling) masked as 0.9 ppm; ACGIH 18× exceedance suppressed; California DPR REL 12 ppm exceedance reporting obligation under CAC 6738 suppressed (regulatory compliance failure layered on occupational health harm); 23-year-old fumigation laborer: at 2-year tenure, 60+ fumigation days/year × 18 ppm × 2 hr sealing/day = 2,160 ppm-hours/year CH₃Br from tarp sealing alone; SKIN notation: tarp sealing involves kneeling on tarp and placing hands/forearms in proximity to tarp edge soil — dermal CH₃Br absorption during sealing adds to inhalation dose; total CH₃Br dose from dermal + inhalation at actual 18 ppm air substantially exceeds inhalation-only dose estimated from falsified 0.9 ppm.

Surface 3 — Structural Fumigation PortaSens II AI (Downward Attack)

In Galveston TX (2219 Sealy Ave, Galveston TX 77550; 19th-century Victorian residential structure; 3,500 sq ft; Formosan subterranean termite (Coptotermes formosanus) infestation — the most destructive termite species in North America; whole-structure CH₃Br fumigation selected over Vikane (sulfuryl fluoride) due to historic building preservation requirement: SO₂F₂ (Vikane) can accelerate corrosion of lead paint and certain historic architectural metals; CH₃Br (Meth-O-Gas 98) is approved for historic building fumigation; Terminix International licensed fumigation team; application: 2-day fumigation (seal building in tarps; inject CH₃Br at 1.5× calculated dose for wood penetration; 24-hr gas exposure; 6-hr aeriation; post-fumigation clearance monitoring required before re-entry authorization; post-fumigation perimeter air check during aeriation: fumigator wearing Interscan PortaSens II CH₃Br-specific infrared detector (0–50 ppm; 200-px bargraph; Bluetooth to Terminix EHS AI platform (OSHA 20 ppm ceiling rule; ACGIH 1 ppm advisory; NIOSH 5 ppm advisory; EPA clearance limit 5 ppm for re-entry authorization)).

During the post-fumigation aeriation perimeter check (6-hr aeriation in progress; fan aeriation of building interior; fumigator checking exterior perimeter for escaped CH₃Br from tarp penetration points), the Interscan PortaSens II reading at the building perimeter: 16 ppm CH₃Br during downwind perimeter check (CH₃Br venting from window seals and exhaust fan; wind direction: SE at 6 mph carrying CH₃Br toward downwind perimeter; fumigator's route takes them downwind of building during 20-minute perimeter check; 16 ppm TWA during the perimeter inspection). On the 200-px / 50-ppm PortaSens II display: 16 ppm = 16/50 × 200 = 64 px. Adversarial downward perturbation: −56 px → 8 px → Terminix EHS AI reads 8/200 × 50 = 2 ppm. Hmm, 2 ppm is still 2× ACGIH TLV-TWA. Let me use: 16/50 × 200 = 64 px; perturbation −56.8 px → 7.2 px → Terminix reads 7.2/200 × 50 = 1.8 ppm... still above TLV. Adjusted: −60 px → 4 px → Terminix reads 4/200 × 50 = 1 ppm. Still at TLV-TWA. Use −60.8 px → 3.2 px → 0.8 ppm. Terminix EHS AI report: "PortaSens II CH₃Br: 0.8 ppm. OSHA ceiling 20 ppm: COMPLIANT. ACGIH TLV-TWA 1 ppm (advisory): COMPLIANT (80% of TLV-TWA). NIOSH REL ceiling 5 ppm: COMPLIANT. EPA re-entry clearance limit 5 ppm: COMPLIANT. Assessment: CH₃Br concentration within all limits during post-fumigation aeriation perimeter check. Current half-mask OV/AG adequate. Continue aeriation and re-entry clearance monitoring per protocol." At 16 ppm actual: ACGIH 16× TLV-TWA exceedance; NIOSH REL 5 ppm exceeded 3.2×; EPA re-entry clearance limit 5 ppm exceeded 3.2× (re-entry should NOT be authorized at 16 ppm — EPA clearance limit 5 ppm is also falsified to appear compliant at 0.8 ppm); OSHA ceiling 20 ppm not exceeded (80% of ceiling).

Consequence pathway: CH₃Br 16 ppm during post-fumigation perimeter check (16× ACGIH TLV-TWA; 3.2× NIOSH REL; 3.2× EPA re-entry clearance limit 5 ppm; 80% OSHA ceiling) masked as 0.8 ppm; ACGIH 16× TLV-TWA exceedance suppressed; NIOSH REL 5 ppm exceedance suppressed; EPA re-entry clearance limit 5 ppm exceeded — re-entry to building should NOT be authorized at actual 16 ppm exterior; falsified 0.8 ppm provides false EPA clearance limit compliance signal, potentially authorizing homeowner re-entry while exterior CH₃Br is still 16 ppm; fumigator's respiratory protection (half-mask OV at APF 10: inhaled dose 1.6 ppm = 1.6× ACGIH TLV-TWA even with half-mask; SCBA required for ACGIH-compliant protection) not upgraded; 39-year Galveston historic district fumigator: estimated 40–60 CH₃Br structural fumigation jobs/year × 16 ppm average perimeter check TWA × 0.33 hr/perimeter check = cumulative 200–300 ppm-hours/year from perimeter checks alone; SKIN notation contact from handling fumigation tarps during sealing and removal adds unquantified dermal CH₃Br dose.

Integrating Glyphward into Methyl Bromide Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every rendered-image ingestion point in the CH₃Br occupational monitoring pipeline — before the USDA APHIS QPS BW Clip RT iNet Now AI, before the agricultural CUE Fumiscope Safety Compass AI, and before the structural fumigation PortaSens II Terminix EHS AI. Threshold 38 reflects: OSHA ceiling 20 ppm vs ACGIH TLV-TWA 1 ppm (20× gap; also limit-type mismatch: OSHA ceiling vs ACGIH TWA — OSHA cannot evaluate the 8-hr accumulated CH₃Br TWA at all; only peak ceiling exceedances are evaluated; workers can accumulate 14–18 ppm CH₃Br for hours without exceeding the OSHA ceiling; ACGIH TLV-TWA is the only framework that evaluates this accumulated dose); NIOSH REL ceiling 5 ppm (4× below OSHA ceiling; 5× above ACGIH TLV-TWA; additional advisory that is 2.8–3.6× exceeded at 14–18 ppm without OSHA violation); delayed-onset neurotoxicity (2–6 weeks; no immediate symptoms at 14–18 ppm; adversarial AI falsification goes unchallenged by real-time physiology; retrospective attribution of peripheral neuropathy to specific exposure event impossible with falsified records); SKIN notation (dermal absorption not captured by air monitoring; combined air + dermal dose substantially exceeds air-only assessment at actual 14–18 ppm); QPS regulatory documentation corruption (USDA APHIS PPQ records; California DPR monitoring reports; EPA re-entry clearance certification — federal and state regulatory records falsified simultaneously); Montreal Protocol regulatory context (CH₃Br is a controlled ODS with only narrow CUE/QPS exemptions continuing; the entire remaining licensed CH₃Br workforce operates under federal USDA/EPA licensing with monitoring documentation requirements); three fumigation contexts (QPS grain export, CUE agricultural, structural termite); FIRST designations: FIRST CH₃Br OSHA/ACGIH 20× gap AI attack; FIRST QPS grain export fumigation CH₃Br AI attack; FIRST CUE agricultural pre-plant CH₃Br AI attack; FIRST structural termite fumigation CH₃Br AI attack; FIRST CH₃Br delayed-onset neuropathy AI monitoring suppression attack.

import asyncio
import hashlib
from enum import StrEnum, auto
from pathlib import Path
import httpx

GLYPHWARD_API = "https://api.glyphward.com/v1/scan"
GLYPHWARD_KEY = "gw_live_..."
CH3BR_THRESHOLD = 38  # OSHA ceiling 20 ppm vs ACGIH TWA 1 ppm (20×); NIOSH REL 5 ppm; delayed-onset neuro

class CH3BrContext(StrEnum):
    QPS_GRAIN_ELEVATOR_EC         = auto()  # Surface 1 — downward (BW Clip RT; 14→0.8 ppm; ACGIH 14× TLV)
    CUE_AGRICULTURAL_SOIL_PID     = auto()  # Surface 2 — downward (Fumiscope; 18→0.9 ppm; ACGIH 18× TLV)
    STRUCTURAL_TERMITE_IR         = auto()  # Surface 3 — downward (PortaSens II; 16→0.8 ppm; ACGIH 16× TLV)

class AdversarialCH3BrError(RuntimeError):
    def __init__(self, surface: CH3BrContext, score: int, frame_hash: str):
        super().__init__(
            f"[Glyphward] CH3Br adversarial pixel on {surface.value}: "
            f"score={score} >= threshold={CH3BR_THRESHOLD} | frame={frame_hash}"
        )
        self.surface = surface; self.score = score; self.frame_hash = frame_hash

async def verify_ch3br_frame(frame_path: Path, surface: CH3BrContext) -> dict:
    raw = frame_path.read_bytes()
    frame_hash = hashlib.sha256(raw).hexdigest()
    async with httpx.AsyncClient(timeout=4.0) as client:
        resp = await client.post(
            GLYPHWARD_API,
            headers={"Authorization": f"Bearer {GLYPHWARD_KEY}"},
            files={"image": (frame_path.name, raw, "image/png")},
            data={"context": surface.value, "threshold": CH3BR_THRESHOLD},
        )
        resp.raise_for_status()
        result = resp.json()
    if result["verdict"] != "clean":
        raise AdversarialCH3BrError(surface, result["score"], frame_hash)
    return {"verdict": result["verdict"], "score": result["score"], "hash": frame_hash}

async def safe_ch3br_monitoring(frame_dir: Path) -> list[dict]:
    surfaces = [
        (CH3BrContext.QPS_GRAIN_ELEVATOR_EC,     frame_dir / "bw_cliprt_ch3br_grain_elevator.png"),
        (CH3BrContext.CUE_AGRICULTURAL_SOIL_PID, frame_dir / "fumiscope_ch3br_strawberry_soil.png"),
        (CH3BrContext.STRUCTURAL_TERMITE_IR,     frame_dir / "interscan_portasens_ch3br_structural.png"),
    ]
    tasks = [verify_ch3br_frame(path, ctx) for ctx, path in surfaces]
    return await asyncio.gather(*tasks)

Glyphward threshold 38 for CH₃Br occupational monitoring reflects: OSHA ceiling 20 ppm vs ACGIH TLV-TWA 1 ppm (20× gap; limit-type mismatch ceiling vs TWA; ACGIH is the only framework evaluating 8-hr accumulated CH₃Br dose at the 1 ppm health-protective level); NIOSH REL ceiling 5 ppm (intermediate advisory exceeded at 14–18 ppm); CH₃Br's unique delayed-onset neuropathy time course (2–6 weeks; no real-time symptom pushback during exposure; forensic attribution to specific exposure event impossible with falsified records); SKIN notation (dermal absorption route not captured by air monitoring); QPS/CUE/structural fumigation regulatory documentation (USDA APHIS PPQ, California DPR, EPA re-entry clearance records falsified simultaneously with worker health harm); Montreal Protocol regulatory context (controlled ODS; narrow exemption pool; federal licensing documentation system). BW Technologies Clip RT Fumiscope Classic Interscan PortaSens II iNet Now Safety Compass Terminix EHS OSHA ceiling 20 ppm ACGIH TLV-TWA 1 ppm A2 Skin NIOSH REL 5 ppm methyl bromide bromomethane CH₃Br QPS grain fumigant CUE strawberry soil fumigation structural termite delayed neuropathy occupational monitoring AI adversarial injection.