Analyzing Network Covert Channels In MalwareSAFE

817 structured cybersecurity skills for AI agents · Mapped to 6 frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND, NIST AI RMF & MITRE F3 (Fight Fraud) · agentskills.io standard · Works with Claude Code, GitHub Copilot, Codex CLI, Cursor, Gemini CLI & 20+ platforms · 29 security domains ·

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: analyzing-network-covert-channels-in-malware
description: Detect and analyze covert communication channels used by malware, including
  DNS tunneling, ICMP exfiltration, steganographic HTTP, and other protocol abuse
  used for C2 and data exfiltration. Use when investigating suspicious DNS/ICMP/HTTP
  traffic patterns, hunting for hidden C2 channels in network captures, or attributing
  exfiltration traffic to a known tunneling toolset.
domain: cybersecurity
subdomain: malware-analysis
tags:
- covert-channels
- dns-tunneling
- icmp-exfiltration
- malware-analysis
- network-forensics
- c2-detection
- data-exfiltration
version: '1.0'
author: mahipal
license: Apache-2.0
d3fend_techniques:
- File Metadata Consistency Validation
- Certificate Analysis
- Application Protocol Command Analysis
- Content Format Conversion
- File Content Analysis
nist_csf:
- DE.AE-02
- RS.AN-03
- ID.RA-01
- DE.CM-01
mitre_attack:
- T1071.001
- T1095
- T1572
- T1001
---
# Analyzing Network Covert Channels in Malware

## Overview

Malware uses covert channels to disguise C2 communication and data exfiltration within legitimate-looking network traffic. DNS tunneling encodes data in DNS queries and responses (used by tools like iodine, dnscat2, and malware families like FrameworkPOS). ICMP tunneling hides data in echo request/reply payloads (icmpsh, ptunnel). HTTP covert channels embed C2 data in headers, cookies, or steganographic images. Protocol abuse exploits allowed protocols to bypass firewalls. DNS tunneling detection achieves 99%+ recall with modern ML-based approaches, though low-throughput exfiltration remains challenging. Palo Alto Unit42 tracked three major DNS tunneling campaigns (TrkCdn, SecShow, Savvy Seahorse) through 2024, showing the technique's continued prevalence.


## When to Use

- When investigating security incidents that require analyzing network covert channels in malware
- When building detection rules or threat hunting queries for this domain
- When SOC analysts need structured procedures for this analysis type
- When validating security monitoring coverage for related attack techniques

## Prerequisites

- Python 3.9+ with `scapy`, `dpkt`, `dnslib`
- Wireshark/tshark for PCAP analysis
- Zeek (formerly Bro) for network monitoring
- DNS query logging infrastructure
- Understanding of DNS, ICMP, HTTP protocols at packet level

## Workflow

### Step 1: DNS Tunneling Detection

```python
#!/usr/bin/env python3
"""Detect DNS tunneling and covert channels in network traffic."""
import sys
import json
import math
from collections import Counter, defaultdict

try:
    from scapy.all import rdpcap, DNS, DNSQR, DNSRR, IP, ICMP
except ImportError:
    print("pip install scapy")
    sys.exit(1)


def entropy(data):
    if not data:
        return 0
    freq = Counter(data)
    length = len(data)
    return -sum((c/length) * math.log2(c/length) for c in freq.values())


def analyze_dns_tunneling(pcap_path):
    """Detect DNS tunneling indicators in PCAP."""
    packets = rdpcap(pcap_path)
    domain_stats = defaultdict(lambda: {
        "queries": 0, "total_qname_len": 0, "subdomain_lengths": [],
        "query_types": Counter(), "unique_subdomains": set(),
    })

    for pkt in packets:
        if pkt.haslayer(DNS) and pkt.haslayer(DNSQR):
            qname = pkt[DNSQR].qname.decode('utf-8', errors='replace').rstrip('.')
            qtype = pkt[DNSQR].qtype

            parts = qname.split('.')
            if len(parts) >= 3:
                base_domain = '.'.join(parts[-2:])
                subdomain = '.'.join(parts[:-2])

                stats = domain_stats[base_domain]
                stats["queries"] += 1
                stats["total_qname_len"] += len(qname)
                stats["subdomain_lengths"].append(len(subdomain))
                stats["query_types"][qtype] += 1
                stats["unique_subdomains"].add(subdomain)

    # Score domains for tunneling indicators
    suspicious = []
    for domain, stats in domain_stats.items():
        if stats["queries"] < 5:
            continue

        avg_subdomain_len = (sum(stats["subdomain_lengths"]) /
                             len(stats["subdomain_lengths"]))
        unique_ratio = len(stats["unique_subdomains"]) / stats["queries"]

        # Calculate subdomain entropy
        all_subdomains = ''.join(stats["unique_subdomains"])
        sub_entropy = entropy(all_subdomains)

        score = 0
        reasons = []

        if avg_subdomain_len > 30:
            score += 30
            reasons.append(f"Long subdomains (avg {avg_subdomain_len:.0f} chars)")
        if unique_ratio > 0.9:
            score += 25
            reasons.append(f"High uniqueness ({unique_ratio:.2%})")
        if sub_entropy > 4.0:
            score += 25
            reasons.append(f"High entropy ({sub_entropy:.2f})")
        if stats["query_types"].get(16, 0) > 10:  # TXT records
            score += 20
            reasons.append(f"Many TXT queries ({stats['query_types'][16]})")

        if score >= 50:
            suspicious.append({
                "domain": domain,
                "score": score,
                "queries": stats["queries"],
                "avg_subdomain_length": round(avg_subdomain_len, 1),
                "unique_subdomains": len(stats["unique_subdomains"]),
                "subdomain_entropy": round(sub_entropy, 2),
                "reasons": reasons,
            })

    return sorted(suspicious, key=lambda x: -x["score"])


def analyze_icmp_tunneling(pcap_path):
    """Detect ICMP tunneling in PCAP."""
    packets = rdpcap(pcap_path)
    icmp_stats = defaultdict(lambda: {"count": 0, "payload_sizes": [], "payloads": []})

    for pkt in packets:
        if pkt.haslayer(ICMP) and pkt.haslayer(IP):
            src = pkt[IP].src
            dst = pkt[IP].dst
            key = f"{src}->{dst}"

            payload = bytes(pkt[ICMP].payload)
            icmp_stats[key]["count"] += 1
            icmp_stats[key]["payload_sizes"].append(len

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

none-observed

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.