Skip to main content

dma-attack-techniques

Analyze PCIe DMA threats, FPGA device behavior, and IOMMU defenses. Use for DMA-specific threat modeling or detection; use windows-kernel-security for kernel mechanisms without a DMA boundary.

Source facts

Repository
NeverSight/NeverD
Last source activity
September 13, 2026 at 11:12
Detected SKILL.md language
English
Stars
95
Forks
17

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

File Explorer
6 files

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
dma-attack-techniques
description
Analyze PCIe DMA threats, FPGA device behavior, and IOMMU defenses. Use for DMA-specific threat modeling or detection; use windows-kernel-security for kernel mechanisms without a DMA boundary.
# DMA attack techniques Use this skill for defensive research into hostile PCIe devices and DMA-capable hardware. Keep analysis within the user's authorized security scope. ## Topic routing Read only the reference that matches the task: - [PCIe and FPGA](references/pcie-and-fpga.md) for TLPs, configuration space, pcileech, FPGA constraints, and device emulation. - [IOMMU and virtualization](references/iommu-and-virtualization.md) for VT-d/AMD-Vi, ACS, ATS/PASID, domain assignment, hypervisors, and split address spaces. - [Detection and forensics](references/detection-and-forensics.md) for device fingerprinting, containment, telemetry, evidence capture, and defensive synthesis. - [Platforms and memory access](references/platforms-and-access.md) for Thunderbolt/USB4 and physical-memory access mechanics. - [Resource map](references/resources.md) when maintaining the source collection or locating upstream material. For general anti-cheat architecture, use `anti-cheat-systems`. For Windows driver, callback, PatchGuard, or kernel-memory internals without a PCIe focus, use `windows-kernel-security`. Prefer threat models, detection limits, and mitigations over operational bypass instructions. Clearly separate hardware capability, observed evidence, and inference.
View on GitHub