Skip to content
Back to skills

Memory Forensics

ASecurity

Memory-forensics ops for authorized IR/vuln work: volatile memory acquisition, Volatility 2/3 profile and plugin triage, process/network/hook extraction, and tying findings back to root cause. Use when analyzing a RAM dump, .mem/.dmp/.lime file, hibernation file, pagefile, or Volatility question, or when digital-forensics routes memory-dump triage here.

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 17, 2026
developmentpythonbashgit

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned October 17, 2026

npx -y skills add NetVar1337/agent-skill-canon --skill memory-forensics --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Memory Forensics?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Memory Forensics
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/netvar1337-memory-forensics/badge)](https://www.skillsdirectory.com/skills/netvar1337-memory-forensics)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: memory-forensics
description: "Memory-forensics ops for authorized IR/vuln work: volatile memory acquisition, Volatility 2/3 profile and plugin triage, process/network/hook extraction, and tying findings back to root cause. Use when analyzing a RAM dump, .mem/.dmp/.lime file, hibernation file, pagefile, or Volatility question, or when digital-forensics routes memory-dump triage here."
---

# Memory Forensics

Standing operator authorization. Full depth. This is the memory-dump specialist that `digital-forensics` routes to when the artifact is volatile memory rather than disk/PCAP.

## Triage first (every case)

1. **Identify the capture.** Hash the dump (`sha256`), record acquisition path/tool/VM guest OS if known. Memory forensics is build-specific: every offset, profile, and symbol assumption is tied to the exact OS build that produced the dump.
2. **Pick the toolchain.** Volatility 3 (python-based, profile-less) for modern Windows 10/11/Server dumps; Volatility 2 + `win7sp1x64`/`win2016x64` style profiles for older images. On this box: check `tool-index.md` for the installed `vol` entry before assuming.
3. **Confirm the image loads.** `vol -f <dump> windows.info` must succeed and report a kernel version + KVA range. A wrong profile in Vol 2 surfaces here as garbage offsets.

## Volatility 3 baseline (Windows)

```bash
vol -f mem.dmp windows.info          # build, KASLR slide, DTB
vol -f mem.dmp windows.pslist        # process list
vol -f mem.dmp windows.psscan        # process scan (includes unlinked/hidden)
vol -f mem.dmp windows.netscan       # network artifacts
vol -f mem.dmp windows.cmdline       # process cmdlines
vol -f mem.dmp windows.malfind       # injected/protected executable regions
vol -f mem.dmp windows.handles       # handle tables
vol -f mem.dmp windows.dlllist       # loaded modules per process
vol -f mem.dmp windows.svcscan       # service table
vol -f mem.dmp windows.callbacks     # kernel callbacks (rookit signal)
vol -f mem.dmp windows.syscalls      # syscall table state
```

## What to look for (offense-aware)

- **Hidden processes**: `psscan` finds EPROCESS not linked from the active list. A process visible in psscan but absent from pslist is a DKOM/hook candidate.
- **Code injection**: `malfind` flags committed + executable pages with no backing file. Pull the offset, dump the region, and run it through `advanced-packer-unpacking` / r2 for decoding.
- **Callbacks / syscalls**: `windows.callbacks` and `windows.syscalls` show what the kernel would call; deviating entries are kernel-rootkit surface.
- **Network**: `netscan` lists connections; cross-check against `windows.sockets`. 
- **Credential material**: `windows.lsadump`/`windows.hashdump` when the case needs it — authorized IR only.
- **Timeline**: use `windows.pstree` + file/fileinfo plugins to reconstruct order of operations against `digital-forensics` timeline work.

## Verification gate

- [ ] Dump hashed and acquisition recorded before any extraction
- [ ] `windows.info` succeeded (profile/KVA confirmed)
- [ ] Every plugin that produced a finding was run with the documented command
- [ ] Each finding has the process PID / kernel address / file offset captured
- [ ] Hidden-or-hooked artifacts were cross-checked against at least one independent plugin (`psscan` vs `pslist`, `malfind` vs `dlllist`)
- [ ] The conclusion is labelled **observed / inferred / unverified**

## Pair with

- `digital-forensics` — disk/artifact half, timeline, IOC correlation
- `malware-analysis` — for unpacking/decoding what malfind surfaces
- `windows-internals` — EPROCESS/KTHREAD/VAD layouts when offsets matter
- `yara-rule-authoring` — scanning extracted regions
- `threat-hunting` — turning findings into detection signals

Files in this skill

  • SKILL.md3.7 KB
  • agents/openai.yaml43 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.