MS-C Mantle — Site-Deployable Adversarial Camouflage

Persistent VIS + IR deception kits for critical sites · Fratres X AI

MS-C Mantle designs manufacturable visual / thermal pattern kits (nets, coatings, tile sheets) that degrade Group 1–2 UAS EO/IR detectors on substations, pads, and pipeline corridors — the passive DENY layer beside active dazzle and transient obscuration.

Status: Preliminary Design + digital prototype · Version: 0.1.0
Repository: https://github.com/Fratres-X-AI/MS-C (private)

Maturity banner: Preliminary Design · Digital PASS (G-DIG) · Physical print/IR OPEN · Field BLOCKED · Tier T1
Stack handoff: Desktop Fratres_Homeland_KillWeb_Handoff/ · Wicked-Web docs/STACK_HANDOFF/
Print path: docs/PRINT_SOW_COUPON_CAMPAIGN.md

Reviewers start here: Executive brief → Stakeholder one-pager → Idea defense → Verification matrix

Conceptual design — NOT validation. Digital surrogate MoE only.
Not a laser. Not an obscurant grenade. Not a fielded stealth system.
Placed by partner survey tools; audited by Wicked Web — not a siting product.

---

Kill-Web slot

Detect → MS-C Mantle (passive persistent deny)
      → MS-V Veil (transient VIS+IR obscure)
      → MPL-D (directed NIR dazzle)
      → MFKS / RADR (kinetic)

Sibling	Role	Persistence
MPL-D	Multi-point laser EO sensor denial	Pulse
MS-V	Multispectral obscurant grenade (Veil)	~120 s
MS-C Mantle (this repo)	Adversarial site camo / pattern kits	Persistent

Tour line: MPL-D dazzles · MS-V obscures · MS-C Mantle denies EO/IR eyes on the site itself.

---

Allowed one-liner

MS-C Mantle is a Preliminary Design + digital prototype for manufacturable VIS/IR adversarial pattern kits that degrade Group 1–2 UAS EO/IR detectors on critical sites — not a fielded stealth system and not a laser effector.

Anything stronger is out of bounds until physical print / IR bench gates close.

---

Program gates

Gate	Status	Meaning
G-DOC	PASS (this package)	Reviewable concept + digital evidence path
G-DIG	PASS (fixture / golden)	Laptop reproduce + unit tests
G-PRINT	OPEN	Physical tile print durability untested
G-IR	OPEN	LWIR chamber / FLIR validation not started
G-FIELD	BLOCKED	No site install authorization

Detail: docs/phase0_gate_status.md

---

Quick start

python -m pip install -e ".[dev]"
make check
# Windows: .\scripts\check.ps1

Optional demo UI:

python -m pip install -e ".[dev,demo]"
python -m demo.app

Optional YOLO GPU optimize: see RUNPOD.md.

---

Digital evidence (honest bounds)

Laptop gate runs a CPU surrogate detector (contrast / edge / thermal proxy) on synthetic site fixtures. Results are planning MoE, not UAS defeat rates.

Metric (digital)	Nominal stack	Adversarial view/light stack
Mean confidence collapse vs uncovered	See golden scorecard	Lower margin by design
Transfer to held-out views	Reported	Not saturated

Golden artifacts: analysis/results/golden/

---

Document map

Doc	Topic
00-executive-brief	One-page concept
STAKEHOLDER_ONE_PAGER	External send sheet
IDEA_DEFENSE_BRIEF	Attack → honest response
SOTA_PASS_1	Public research positioning
CONOPS_HOMELAND	Site install / refresh
KILLWEB_INTERFACE_NOTE	Passive layer ICD
LAYERED_STACK	Mantle ≠ Veil ≠ MPL-D
REQUIREMENTS	REQ list
RISK_REGISTER	Residual risks
EXPORT_CONTROL_SCREENING	Screening stub

---

Layout

Path	Role
models/	Scenes, materials, dual-band renderer, optimizer, detectors
sim/	Reproduce, optimize CLI, scorecard, kit export
data/	Seedable synthetic site fixtures
rtm/	Verification matrix
analysis/	Golden evidence + run outputs
demo/	Gradio tour UI
tests/	Unit + invariant tests

---

Non-goals (v1)

• No laser / dazzler physics → MPL-D
• No obscurant plume M&S → MS-V
• No 3DGS full-site reconstruction (Phase-1 stretch)
• No physical IR chamber claims; BOM is planning-only

---

Fratres X AI | Defense Projects — Prototype Documentation · MS-C Mantle
All digital MoE outputs are notional; not authorization to procure, manufacture, export, or field.