Collecting Volatile Evidence from Compromised Hosts
SkillDocs & knowledgeA forensic incident-response skill that guides an AI agent through collecting volatile evidence from a compromised host.
Use Collecting Volatile Evidence from Compromised Hosts in Claude, ChatGPT or Ahel Desktop
Free. Sign in, add Collecting Volatile Evidence from Compromised Hosts and connect your AI. About a minute.
Also: Claude Code · Cursor · Codex
Then ask your AI: use the Collecting Volatile Evidence from Compromised Hosts skill
Details
Instructions available. Your AI can read the instructions. Execution depends on the setup they require.
Account requirements not reviewed. Check the skill instructions before use; ahel provides instructions and does not run this skill.
No other account needed.
Have a forensic collection toolkit with trusted tools available on USB media or a network share, and verify its integrity before use.
What your AI can do with it
- Dump system memory using WinPmem, LiME, /proc/kcore, or osxpmem depending
- Capture active network connections and running processes from the compromised host
- Record logged-in users and system configuration as part of evidence collection
- Log timestamps, collector identity, and SHA-256 hashes for a documented chain of custody
- Verify the integrity of the forensic toolkit against known-good hashes before collection
- Create timestamped evidence output directories with a collection log
Getting started
- Have a forensic collection toolkit with trusted tools available on USB media or a network share, and verify its integrity before use.
- Ensure memory acquisition tools are available: WinPmem for Windows, LiME for Linux, and osxpmem for macOS.
- Prepare chain of custody documentation forms and secure evidence storage with integrity verification.
- Obtain authorization to collect evidence, including legal or HR approval for insider cases.
- Add the skill to your agent and run it before isolating, shutting down, or remediating the compromised host.
What this skill tells your AI
The instructions your AI receives, as published by mukul975/anthropic-cybersecurity-skills in skills/collecting-volatile-evidence-from-compromised-host/SKILL.md and read by ahel’s review.
When to Use
- Security incident confirmed and compromised host identified
- Before system isolation, shutdown, or remediation begins
- Memory-resident malware suspected (fileless attacks)
- Need to capture network connections, running processes, and system state
- Legal proceedings may require forensic evidence preservation
- Incident requires root cause analysis with volatile data
Prerequisites
- Forensic collection toolkit on USB or network share (trusted tools)
- WinPmem/LiME for memory acquisition
- Write-blocker or forensic workstation for disk imaging
- Chain of custody documentation forms
- Secure evidence storage with integrity verification
- Authorization to collect evidence (legal/HR approval for insider cases)
Workflow
Step 1: Prepare Collection Environment
# Mount forensic USB toolkit (do NOT install tools on compromised system)
# Verify toolkit integrity
sha256sum /mnt/forensic_usb/tools/* > /tmp/toolkit_hashes.txt
diff /mnt/forensic_usb/tools/known_good_hashes.txt /tmp/toolkit_hashes.txt
# Create evidence output directory with timestamps
EVIDENCE_DIR="/mnt/evidence/$(hostname)_$(date +%Y%m%d_%H%M%S)"
mkdir -p "$EVIDENCE_DIR"
echo "Collection started: $(date -u)" > "$EVIDENCE_DIR/collection_log.txt"
echo "Collector: $(whoami)" >> "$EVIDENCE_DIR/collection_log.txt"
echo "System: $(hostname)" >> "$EVIDENCE_DIR/collection_log.txt"
Step 2: Capture System Memory (Highest Volatility)
# Windows - WinPmem memory acquisition
winpmem_mini_x64.exe "$EVIDENCE_DIR\memdump_$(hostname).raw"
# Linux - LiME kernel module for memory acquisition
insmod /mnt/forensic_usb/lime.ko "path=$EVIDENCE_DIR/memdump_$(hostname).lime format=lime"
# Linux - Alternative using /proc/kcore
dd if=/proc/kcore of="$EVIDENCE_DIR/kcore_dump.raw" bs=1M
# macOS - osxpmem
osxpmem -o "$EVIDENCE_DIR/memdump_$(hostname).aff4"
# Hash the memory dump immediately
sha256sum "$EVIDENCE_DIR/memdump_"* > "$EVIDENCE_DIR/memory_hash.sha256"
Step 3: Capture Network State
# Active network connections
# Windows
netstat -anob > "$EVIDENCE_DIR/netstat_connections.txt" 2>&1
Get-NetTCPConnection | Export-Csv "$EVIDENCE_DIR/tcp_connections.csv" -NoTypeInformation
Get-NetUDPEndpoint | Export-Csv "$EVIDENCE_DIR/udp_endpoints.csv" -NoTypeInformation
# Linux
ss -tulnp > "$EVIDENCE_DIR/socket_stats.txt"
netstat -anp > "$EVIDENCE_DIR/netstat_all.txt" 2>/dev/null
cat /proc/net/tcp > "$EVIDENCE_DIR/proc_net_tcp.txt"
cat /proc/net/udp > "$EVIDENCE_DIR/proc_net_udp.txt"
# ARP cache
arp -a > "$EVIDENCE_DIR/arp_cache.txt"
# Routing table
route print > "$EVIDENCE_DIR/routing_table.txt" # Windows
ip route show > "$EVIDENCE_DIR/routing_table.txt" # Linux
# DNS cache
ipconfig /displaydns > "$EVIDENCE_DIR/dns_cache.txt" # Windows
# Linux: varies by resolver, check systemd-resolve or nscd
systemd-resolve --statistics > "$EVIDENCE_DIR/dns_stats.txt" 2>/dev/null
# Active firewall rules
netsh advfirewall show allprofiles > "$EVIDENCE_DIR/firewall_rules.txt" # Windows
iptables -L -n -v > "$EVIDENCE_DIR/iptables_rules.txt" # Linux
Step 4: Capture Running Processes
# Windows - Detailed process list
tasklist /V /FO CSV > "$EVIDENCE_DIR/process_list_verbose.csv"
wmic process list full > "$EVIDENCE_DIR/wmic_process_full.txt"
Get-Process | Select-Object Id,ProcessName,Path,StartTime,CPU,WorkingSet |
Export-Csv "$EVIDENCE_DIR/ps_processes.csv" -NoTypeInformation
# Windows - Process with command line and parent
wmic process get ProcessId,Name,CommandLine,ParentProcessId,ExecutablePath /FORMAT:CSV > \
"$EVIDENCE_DIR/process_commandlines.csv"
# Linux - Full process tree
ps auxwwf > "$EVIDENCE_DIR/process_tree.txt"
ps -eo pid,ppid,user,args --forest > "$EVIDENCE_DIR/process_forest.txt"
cat /proc/*/cmdline 2>/dev/null | tr '\0' ' ' > "$EVIDENCE_DIR/proc_cmdline_all.txt"
# Process modules/DLLs loaded
# Windows
listdlls.exe -accepteula > "$EVIDENCE_DIR/loaded_dlls.txt"
# Linux
for pid in $(ls /proc/ | grep -E '^[0-9]+$'); do
echo "=== PID $pid ===" >> "$EVIDENCE_DIR/proc_maps.txt"
cat "/proc/$pid/maps" 2>/dev/null >> "$EVIDENCE_DIR/proc_maps.txt"
done
# Open file handles
handle.exe -accepteula > "$EVIDENCE_DIR/open_handles.txt" # Windows (Sysinternals)
lsof > "$EVIDENCE_DIR/open_files.txt" # Linux
Step 5: Capture Logged-in Users and Sessions
# Windows
query user > "$EVIDENCE_DIR/logged_in_users.txt"
query session > "$EVIDENCE_DIR/active_sessions.txt"
net session > "$EVIDENCE_DIR/net_sessions.txt" 2>&1
net use > "$EVIDENCE_DIR/mapped_drives.txt" 2>&1
# Linux
who > "$EVIDENCE_DIR/who_output.txt"
w > "$EVIDENCE_DIR/w_output.txt"
last -50 > "$EVIDENCE_DIR/last_logins.txt"
lastlog > "$EVIDENCE_DIR/lastlog.txt"
cat /var/log/auth.log | tail -200 > "$EVIDENCE_DIR/recent_auth.txt" 2>/dev/null
Step 6: Capture System Configuration State
# System time (critical for timeline)
date -u > "$EVIDENCE_DIR/system_time_utc.txt"
w32tm /query /status > "$EVIDENCE_DIR/ntp_status.txt" # Windows
ntpq -p > "$EVIDENCE_DIR/ntp_status.txt" # Linux
# Environment variables
set > "$EVIDENCE_DIR/environment_vars.txt" # Windows
env > "$EVIDENCE_DIR/environment_vars.txt" # Linux
# Scheduled tasks / Cron jobs
schtasks /query /fo CSV /v > "$EVIDENCE_DIR/scheduled_tasks.csv" # Windows
crontab -l > "$EVIDENCE_DIR/crontab_current.txt" 2>/dev/null # Linux
ls -la /etc/cron.* > "$EVIDENCE_DIR/cron_dirs.txt" 2>/dev/null
# Services
sc queryex type=service state=all > "$EVIDENCE_DIR/services_all.txt" # Windows
systemctl list-units --type=service --all > "$EVIDENCE_DIR/systemd_services.txt" # Linux
# Windows Registry - key autostart locations
reg export "HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Run" "$EVIDENCE_DIR/reg_run_hklm.reg" /y
reg export "HKCU\SOFTWARE\Microsoft\Windows\CurrentVersion\Run" "$EVIDENCE_DIR/reg_run_hkcu.reg" /y
reg export "HKLM\SYSTEM\CurrentControlSet\Services" "$EVIDENCE_DIR/reg_services.reg" /y
Step 7: Hash All Evidence and Document Chain of Custody
# Generate SHA256 hashes for all collected evidence
cd "$EVIDENCE_DIR"
sha256sum * > evidence_manifest.sha256
# Create chain of custody record
cat > "$EVIDENCE_DIR/chain_of_custody.txt" << EOF
CHAIN OF CUSTODY RECORD
========================
Case ID: IR-YYYY-NNN
Collection Date: $(date -u)
Collected By: $(whoami)
System: $(hostname)
System IP: $(hostname -I 2>/dev/null || ipconfig | grep IPv4)
Collection Method: Live forensic collection via trusted USB toolkit
Evidence Items:
$(ls -la "$EVIDENCE_DIR/" | grep -v chain_of_custody)
SHA256 Manifest: evidence_manifest.sha256
Transfer: [TO BE COMPLETED]
Storage Location: [TO BE COMPLETED]
EOF
Key Concepts
| Concept | Description |
|---|---|
| Order of Volatility | RFC 3227 - Collect most volatile data first: registers > cache > memory > disk |
| Live Forensics | Collecting evidence from a running system before shutdown |
| Chain of Custody | Documentation tracking evidence handling from collection to court |
| Forensic Soundness | Ensuring evidence collection doesn't alter the original evidence |
| Trusted Tools | Using verified tools from external media, not from the compromised system |
| Evidence Integrity | SHA256 hashing of all evidence immediately after collection |
| Locard's Exchange Principle | Every contact leaves a trace - minimize investigator artifacts |
Tools & Systems
| Tool | Purpose |
|---|---|
| WinPmem | Windows memory acquisition |
| LiME (Linux Memory Extractor) | Linux kernel memory acquisition |
| Sysinternals Suite | Process, handle, and DLL analysis (Windows) |
| Velociraptor | Remote forensic collection at scale |
| KAPE (Kroll Artifact Parser) | Automated artifact collection on Windows |
| CyLR | Cross-platform live response collection |
| GRR Rapid Response | Remote live forensics framework |
Common Scenarios
- Fileless Malware Attack: PowerShell-based attack with no files on disk. Memory dump is critical evidence containing the malicious scripts.
- Active C2 Session: Attacker has live connection. Network connections and process data reveal C2 infrastructure.
- Insider Data Theft: Employee copying files. Process list, mapped drives, and network connections show exfiltration activity.
- Compromised Web Server: Web shell detected. Memory may contain additional backdoors not yet written to disk.
- Lateral Movement in Progress: Attacker moving between systems. Authentication tokens and network sessions in memory reveal scope.
Output Format
- Memory dump file (.raw or .lime format) with SHA256 hash
- Network state captures (connections, ARP, DNS, routes)
- Process listings with command lines and parent processes
- User session and authentication data
- System configuration snapshots
- Evidence manifest with SHA256 checksums
- Chain of custody documentation
Signals
- GitHub stars
- 34k
- Forks
- 4k
- Last commit
- Aug 2026
ahel review
K1binfo
installs-packages (in scripts/process.py)
Automated review, not a security audit. Ruleset v1+k2.
Questions
- When should this skill be used?
- When a security incident is confirmed and a compromised host is identified, before isolation, shutdown, or remediation begins. It also applies when fileless or memory-resident malware is suspected, root cause analysis is needed, or evidence must hold up in legal proceedings.
- Why collect volatile evidence before shutting down or remediating?
- Volatile data such as memory contents, network connections, and running processes are lost when a machine is isolated, shut down, or remediated. The skill preserves them first, in order of volatility.
Advanced
- Item type
- skill
- Key
collecting-volatile-evidence-from-compromised-host-mukul975- Source
- github.com/mukul975/anthropic-cybersecurity-skills
github.com/mukul975/anthropic-cybersecurity-skills
Related picks
Skill · nvidia
The pick for Infrahttp-to-https
Skill · thedaviddias
The pick for Infrasecrets-exposure-review
Skill · naodeng
The pick for Secretssecrets-with-git-crypt
Skill · derailed-dash
The pick for Secretshandoff
Skill · mattpocock
More in Docs & knowledgecanvas-design
Skill · anthropics
More in Docs & knowledge