Agent-Skills.md

Agent Skills: Detecting Ransomware Precursors in Network Traffic

>

UncategorizedID: plurigrid/asi/detecting-ransomware-precursors-in-network

Repository

plurigrid/asi
plurigridLicense: MIT
165

Install this agent skill to your local

pnpm dlx add-skill https://github.com/plurigrid/asi/tree/HEAD/plugins/asi/skills/detecting-ransomware-precursors-in-network

Skill Files

Browse the full folder contents for detecting-ransomware-precursors-in-network.

Download Skill

Loading file tree…

plugins/asi/skills/detecting-ransomware-precursors-in-network/SKILL.md

Skill Metadata

Name
detecting-ransomware-precursors-in-network
Description
>

Detecting Ransomware Precursors in Network Traffic

When to Use

  • Building detection rules for pre-ransomware network activity (the average time from Cobalt Strike deployment to encryption is 17 minutes)
  • Monitoring for initial access broker (IAB) indicators that precede ransomware deployment
  • Creating SIEM correlation rules that chain multiple precursor events into high-confidence alerts
  • Tuning network detection systems to distinguish ransomware staging from normal administrative activity
  • Investigating suspicious network patterns that may indicate ransomware operators have established a foothold

Do not use for post-encryption response (see recovering-from-ransomware-attack). This skill focuses on the pre-encryption detection window where containment can prevent data loss.

Prerequisites

  • Network detection platform (Zeek/Bro, Suricata, or Arkime/Moloch) deployed on network TAP or SPAN ports
  • SIEM platform (Splunk, Elastic Security, Microsoft Sentinel, or QRadar) ingesting network logs
  • Threat intelligence feeds covering ransomware IOCs (CISA, abuse.ch, OTX, MISP)
  • Network flow data (NetFlow/IPFIX) from core routers and firewalls
  • DNS query logging from internal resolvers
  • Full packet capture capability for incident investigation

Workflow

Step 1: Identify Ransomware Kill Chain Phases in Network Traffic

Map network-observable indicators to each pre-encryption phase:

| Kill Chain Phase | Network Indicators | Detection Source | |------------------|--------------------|------------------| | Initial Access | RDP brute force, VPN credential stuffing, phishing callback | Firewall logs, IDS, proxy logs | | C2 Establishment | Cobalt Strike beacons (HTTPS/DNS), Sliver/Brute Ratel callbacks | Zeek SSL/HTTP logs, DNS logs | | Credential Harvesting | NTLM relay, Kerberoasting, DCSync traffic | Zeek Kerberos/NTLM logs, DC logs | | Reconnaissance | Internal port scanning, AD enumeration (LDAP/SMB) | Zeek conn.log, flow data | | Lateral Movement | PsExec/WMI/WinRM traffic, RDP pivoting, SMB file copies | Zeek SMB/DCE-RPC logs | | Staging | Data aggregation, archive creation, cloud upload prep | Proxy logs, DNS logs, DLP |

Step 2: Deploy Network Detection Rules

Suricata rules for common ransomware precursors:

# Cobalt Strike default HTTPS beacon profile detection
alert tls $HOME_NET any -> $EXTERNAL_NET any (msg:"RANSOMWARE PRECURSOR - Cobalt Strike Default TLS Certificate"; tls.cert_subject; content:"Major Cobalt Strike"; sid:3000001; rev:1;)

# Cobalt Strike DNS beacon
alert dns $HOME_NET any -> any 53 (msg:"RANSOMWARE PRECURSOR - Cobalt Strike DNS Beacon Pattern"; dns.query; pcre:"/^[a-z0-9]{3}\.[a-z]{4,8}\./"; threshold:type both, track by_src, count 50, seconds 60; sid:3000002; rev:1;)

# Mimikatz network signature (DCSync - DRS GetNCChanges)
alert tcp $HOME_NET any -> $HOME_NET 135 (msg:"RANSOMWARE PRECURSOR - Possible DCSync/Mimikatz"; content:"|05 00 0b|"; offset:0; depth:3; content:"|e3 51 4d 2b 4b 47 15 d2|"; sid:3000003; rev:1;)

# Internal network scanning (many connections, few bytes)
alert tcp $HOME_NET any -> $HOME_NET any (msg:"RANSOMWARE PRECURSOR - Internal Port Scan"; flags:S; threshold:type both, track by_src, count 100, seconds 10; sid:3000004; rev:1;)

# PsExec service installation over SMB
alert tcp $HOME_NET any -> $HOME_NET 445 (msg:"RANSOMWARE PRECURSOR - PsExec Service Install"; content:"|ff|SMB"; content:"PSEXESVC"; nocase; sid:3000005; rev:1;)

# RDP brute force from internal host (lateral movement)
alert tcp $HOME_NET any -> $HOME_NET 3389 (msg:"RANSOMWARE PRECURSOR - Internal RDP Brute Force"; flow:to_server,established; threshold:type both, track by_src, count 20, seconds 60; sid:3000006; rev:1;)

# Large SMB file transfer (data staging)
alert tcp $HOME_NET any -> $HOME_NET 445 (msg:"RANSOMWARE PRECURSOR - Large SMB Transfer Possible Staging"; flow:to_server,established; dsize:>60000; threshold:type both, track by_src, count 100, seconds 300; sid:3000007; rev:1;)

Zeek scripts for behavioral detection:

# detect_ransomware_precursors.zeek
# Detect high volume of failed SMB connections (credential testing)

@load base/protocols/smb

module RansomwarePrecursor;

export {
    redef enum Notice::Type += {
        SMB_Brute_Force,
        Suspicious_Internal_Scan,
        Excessive_DNS_Queries,
        SMB_Admin_Share_Access,
    };

    const smb_fail_threshold = 10 &redef;
    const scan_threshold = 50 &redef;
    const dns_query_threshold = 200 &redef;
}

global smb_fail_count: table[addr] of count &default=0 &create_expire=5min;
global conn_count: table[addr] of set[addr] &create_expire=1min;

event smb2_message(c: connection, hdr: SMB2::Header, is_orig: bool) {
    if (hdr$status != 0) {
        ++smb_fail_count[c$id$orig_h];
        if (smb_fail_count[c$id$orig_h] >= smb_fail_threshold) {
            NOTICE([$note=SMB_Brute_Force,
                    $msg=fmt("Host %s has %d failed SMB attempts", c$id$orig_h, smb_fail_count[c$id$orig_h]),
                    $src=c$id$orig_h,
                    $identifier=cat(c$id$orig_h)]);
        }
    }
}

event new_connection(c: connection) {
    if (c$id$orig_h in Site::local_nets && c$id$resp_h in Site::local_nets) {
        if (c$id$orig_h !in conn_count)
            conn_count[c$id$orig_h] = set();
        add conn_count[c$id$orig_h][c$id$resp_h];
        if (|conn_count[c$id$orig_h]| >= scan_threshold) {
            NOTICE([$note=Suspicious_Internal_Scan,
                    $msg=fmt("Host %s connected to %d internal hosts in 1 min", c$id$orig_h, |conn_count[c$id$orig_h]|),
                    $src=c$id$orig_h,
                    $identifier=cat(c$id$orig_h)]);
        }
    }
}

Step 3: Create SIEM Correlation Rules

Splunk correlation for ransomware precursor chain:

| tstats count FROM datamodel=Network_Traffic
  WHERE earliest=-24h All_Traffic.dest_port IN (445, 135, 139, 3389, 5985, 5986)
    AND All_Traffic.src_ip IN 10.0.0.0/8
    AND All_Traffic.dest_ip IN 10.0.0.0/8
  BY All_Traffic.src_ip, All_Traffic.dest_port, _time span=1h
| stats dc(All_Traffic.dest_port) as port_count,
        values(All_Traffic.dest_port) as ports,
        count as total_conns
  BY All_Traffic.src_ip
| where port_count >= 3 AND total_conns > 50
| rename All_Traffic.src_ip as src_ip
| lookup threat_intel_ioc ip as src_ip OUTPUT threat_type
| eval risk_score = case(
    port_count >= 5 AND total_conns > 200, "CRITICAL",
    port_count >= 3 AND total_conns > 50, "HIGH",
    1=1, "MEDIUM")
| table src_ip, ports, port_count, total_conns, risk_score, threat_type

Microsoft Sentinel KQL - Ransomware precursor correlation:

let timeframe = 24h;
let RDPBruteForce = SecurityEvent
| where TimeGenerated > ago(timeframe)
| where EventID == 4625
| where LogonType == 10
| summarize FailedRDP = count() by TargetAccount, IpAddress, bin(TimeGenerated, 1h)
| where FailedRDP > 10;
let SuspiciousSMB = SecurityEvent
| where TimeGenerated > ago(timeframe)
| where EventID == 5145
| where ShareName has "ADMIN$" or ShareName has "C$" or ShareName has "IPC$"
| summarize AdminShareAccess = count() by SubjectUserName, IpAddress, bin(TimeGenerated, 1h)
| where AdminShareAccess > 5;
let ServiceInstalls = SecurityEvent
| where TimeGenerated > ago(timeframe)
| where EventID == 7045
| where ServiceName has_any ("PSEXESVC", "meterpreter", "beacon");
RDPBruteForce
| join kind=inner SuspiciousSMB on IpAddress
| project TimeGenerated, IpAddress, TargetAccount, FailedRDP, SubjectUserName, AdminShareAccess
| extend AlertTitle = "Ransomware Precursor: RDP Brute Force + Admin Share Access"

Step 4: Integrate Threat Intelligence

Configure automated IOC feeds for known ransomware infrastructure:

# Download and update ransomware C2 blocklists
# abuse.ch Feodo Tracker (Cobalt Strike, TrickBot, BazarLoader C2s)
curl -s https://feodotracker.abuse.ch/downloads/ipblocklist.csv | \
  grep -v "^#" | cut -d, -f2 > /opt/threat-intel/feodo_ips.txt

# abuse.ch URLhaus (malware distribution URLs)
curl -s https://urlhaus.abuse.ch/downloads/csv_recent/ | \
  grep -v "^#" | cut -d, -f3 > /opt/threat-intel/urlhaus_urls.txt

# abuse.ch ThreatFox (ransomware IOCs)
curl -s https://threatfox.abuse.ch/export/csv/recent/ | \
  grep -i "ransomware" | cut -d, -f3 > /opt/threat-intel/ransomware_iocs.txt

# CISA Known Exploited Vulnerabilities (initial access vectors)
curl -s https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json | \
  python3 -c "import json,sys; data=json.load(sys.stdin); [print(v['cveID'],v['vendorProject'],v['product']) for v in data['vulnerabilities'] if 'ransomware' in v.get('knownRansomwareCampaignUse','').lower()]"

Step 5: Establish Alert Triage and Escalation

Define triage procedures based on precursor confidence level:

| Alert Type | Confidence | Response Time | Action | |------------|-----------|---------------|--------| | Confirmed Cobalt Strike beacon | High | 15 minutes | Isolate host immediately, trigger IR | | DCSync/Kerberoasting from non-DC | High | 15 minutes | Disable account, isolate host, trigger IR | | Internal port scan + admin share access | Medium-High | 30 minutes | Investigate source host, check EDR telemetry | | RDP brute force from internal host | Medium | 1 hour | Verify if legitimate admin activity, check host | | Unusual DNS query volume | Low-Medium | 4 hours | Check for DNS tunneling, correlate with other alerts |

Key Concepts

| Term | Definition | |------|------------| | Ransomware Precursor | Network activity that precedes ransomware encryption, including C2 communication, lateral movement, and data staging | | Dwell Time | Time between initial compromise and ransomware deployment, averaging 21 days but sometimes as short as 17 minutes | | Initial Access Broker (IAB) | Threat actors who sell compromised network access to ransomware operators on dark web markets | | Beaconing | Periodic C2 callbacks from implants (Cobalt Strike, Sliver) that can be detected by analyzing connection timing patterns | | Kerberoasting | Credential harvesting technique requesting Kerberos service tickets for offline cracking, detectable via unusual TGS-REQ patterns | | DCSync | Technique using Directory Replication Service to extract password hashes from domain controllers, critical ransomware precursor |

Tools & Systems

  • Zeek (formerly Bro): Network analysis framework generating structured logs for SMB, Kerberos, DNS, HTTP, and TLS connections
  • Suricata: High-performance IDS/IPS with protocol analysis and multi-threading support for ransomware signature detection
  • Arkime (formerly Moloch): Full packet capture and search platform for deep forensic investigation of network events
  • RITA (Real Intelligence Threat Analytics): Open-source tool for detecting beaconing, DNS tunneling, and long connections in Zeek logs
  • AC-Hunter: Network threat hunting platform from Active Countermeasures for beacon detection and C2 identification

Common Scenarios

Scenario: Detecting LockBit Precursors in a Manufacturing Network

Context: A manufacturing company's SOC receives an alert for unusual SMB traffic from a workstation (10.1.5.42) in the engineering department. The workstation connected to 47 internal hosts on port 445 within 5 minutes at 2:00 AM.

Approach:

  1. Zeek conn.log analysis shows 10.1.5.42 initiated connections to 47 unique internal IPs on port 445, 135, and 3389 between 01:55-02:05
  2. Zeek ssl.log reveals an outbound HTTPS connection to 185.x.x.x every 60 seconds with consistent 48-byte payloads (Cobalt Strike beacon pattern)
  3. RITA beacon analysis confirms high beacon score (0.96) for the external IP with 60-second jitter
  4. Zeek kerberos.log shows TGS-REQ for multiple SPN accounts from 10.1.5.42 (Kerberoasting)
  5. SMB tree_connect events show access to ADMIN$ shares on 12 hosts (lateral movement staging)
  6. Containment: Host isolated, credentials for engineering user reset, blocking rule for C2 IP deployed
  7. Full IR initiated before ransomware deployment could begin

Pitfalls:

  • Dismissing internal port scans as vulnerability scanner activity without verifying the source is an authorized scanner
  • Not correlating individual low-severity alerts (DNS anomaly + SMB access + failed logins) into a high-severity chain
  • Setting detection thresholds too high to avoid false positives, missing low-and-slow reconnaissance
  • Ignoring encrypted traffic analysis (JA3/JA4 fingerprinting) that can identify Cobalt Strike even in TLS tunnels

Output Format

## Ransomware Precursor Detection Alert

**Alert ID**: [SIEM-generated ID]
**Detection Time**: [Timestamp]
**Source Host**: [IP / Hostname]
**Confidence**: [High / Medium / Low]
**Kill Chain Phase**: [Initial Access / C2 / Credential Harvest / Recon / Lateral Movement / Staging]

### Indicators Detected
| Indicator | Source | Detail | MITRE ATT&CK |
|-----------|--------|--------|--------------|
| [Type] | [Zeek/Suricata/SIEM] | [Description] | [T-ID] |

### Correlation Chain
1. [Timestamp] - [Event 1]
2. [Timestamp] - [Event 2]
3. [Timestamp] - [Event 3]

### Recommended Actions
- [ ] Isolate source host from network
- [ ] Check EDR telemetry for host-based indicators
- [ ] Reset credentials for affected user accounts
- [ ] Block identified C2 infrastructure
- [ ] Escalate to incident response team