Anthropic-Cybersecurity-Skills/skills/implementing-soar-automation-with-phantom/SKILL.md

---
name: implementing-soar-automation-with-phantom
description: >
  Implements Security Orchestration, Automation, and Response (SOAR) workflows using Splunk SOAR
  (formerly Phantom) to automate alert triage, IOC enrichment, containment actions, and incident
  response playbooks. Use when SOC teams need to reduce manual analyst work, standardize response
  procedures, or integrate multiple security tools into automated workflows.
domain: cybersecurity
subdomain: soc-operations
tags: [soc, soar, phantom, splunk-soar, automation, playbook, orchestration, incident-response]
version: "1.0"
author: mahipal
license: MIT
---
# Implementing SOAR Automation with Phantom

## When to Use

Use this skill when:
- SOC teams need to automate repetitive triage and enrichment tasks for high-volume alerts
- Manual response times exceed SLA requirements and automation can reduce MTTR
- Multiple security tools (SIEM, EDR, firewall, TIP) need orchestrated response actions
- Playbook standardization is required to ensure consistent analyst response across shifts

**Do not use** for fully autonomous containment without human approval gates — always include analyst decision points for high-impact actions like account disabling or host isolation.

## Prerequisites

- Splunk SOAR (Phantom) 6.x+ deployed with web interface access
- App connectors configured: VirusTotal, CrowdStrike, ServiceNow, Active Directory, Splunk ES
- Splunk ES integration for ingesting notable events as SOAR events
- API credentials for each integrated tool stored in SOAR asset configuration
- Python knowledge for custom playbook actions

## Workflow

### Step 1: Configure Asset Connections

Set up integrations with security tools via SOAR Apps:

**VirusTotal Asset Configuration:**
```json
{
  "app": "VirusTotal v3",
  "asset_name": "virustotal_prod",
  "configuration": {
    "api_key": "YOUR_VT_API_KEY",
    "rate_limit": true,
    "max_requests_per_minute": 4
  },
  "product_vendor": "VirusTotal",
  "product_name": "VirusTotal"
}
```

**CrowdStrike Falcon Asset:**
```json
{
  "app": "CrowdStrike Falcon",
  "asset_name": "crowdstrike_prod",
  "configuration": {
    "client_id": "CS_CLIENT_ID",
    "client_secret": "CS_CLIENT_SECRET",
    "base_url": "https://api.crowdstrike.com"
  }
}
```

**Active Directory Asset:**
```json
{
  "app": "Active Directory",
  "asset_name": "ad_prod",
  "configuration": {
    "server": "dc01.company.com",
    "username": "soar_service@company.com",
    "password": "SERVICE_ACCOUNT_PASSWORD",
    "ssl": true
  }
}
```

### Step 2: Build Phishing Triage Playbook

Create an automated phishing response playbook in Python (Phantom playbook format):

```python
"""
Phishing Triage Automation Playbook
Trigger: New phishing email reported via Splunk ES notable or email ingestion
"""

import phantom.rules as phantom
import json

def on_start(container):
    # Extract artifacts (URLs, file hashes, sender) from the container
    artifacts = phantom.get_artifacts(container_id=container["id"])

    for artifact in artifacts:
        artifact_type = artifact.get("cef", {}).get("type", "")

        if artifact_type == "url":
            phantom.act("url reputation", targets=artifact,
                        assets=["virustotal_prod"],
                        callback=url_reputation_callback,
                        name="url_reputation")

        elif artifact_type == "hash":
            phantom.act("file reputation", targets=artifact,
                        assets=["virustotal_prod"],
                        callback=hash_reputation_callback,
                        name="file_reputation")

        elif artifact_type == "ip":
            phantom.act("ip reputation", targets=artifact,
                        assets=["virustotal_prod"],
                        callback=ip_reputation_callback,
                        name="ip_reputation")

def url_reputation_callback(action, success, container, results, handle):
    if not success:
        phantom.comment(container, "URL reputation check failed")
        return

    for result in results:
        data = result.get("data", [{}])[0]
        malicious_count = data.get("summary", {}).get("malicious", 0)
        total_engines = data.get("summary", {}).get("total_engines", 0)

        if malicious_count > 5:
            # High confidence malicious — auto-block and escalate
            phantom.act("block url", targets=result,
                        assets=["palo_alto_prod"],
                        name="block_malicious_url")

            phantom.set_severity(container, "high")
            phantom.set_status(container, "open")
            phantom.comment(container,
                f"URL flagged by {malicious_count}/{total_engines} engines. "
                f"Blocked on firewall. Escalating to Tier 2.")

            # Create ServiceNow ticket
            phantom.act("create ticket", targets=container,
                        assets=["servicenow_prod"],
                        parameters=[{
                            "short_description": f"Phishing - Malicious URL detected",
                            "urgency": "2",
                            "impact": "2"
                        }],
                        name="create_incident_ticket")

        elif malicious_count > 0:
            # Medium confidence — request analyst review
            phantom.promote(container, template="Phishing Investigation")
            phantom.comment(container,
                f"URL flagged by {malicious_count}/{total_engines} engines. "
                f"Requires analyst review.")

        else:
            # Clean — close with comment
            phantom.set_status(container, "closed")
            phantom.comment(container,
                f"URL clean: 0/{total_engines} engines flagged. Auto-closed.")

def hash_reputation_callback(action, success, container, results, handle):
    if not success:
        return

    for result in results:
        data = result.get("data", [{}])[0]
        positives = data.get("summary", {}).get("positives", 0)

        if positives > 10:
            # Known malware — quarantine and block
            phantom.act("quarantine device", targets=result,
                        assets=["crowdstrike_prod"],
                        name="isolate_endpoint")
            phantom.set_severity(container, "high")

def ip_reputation_callback(action, success, container, results, handle):
    if not success:
        return

    for result in results:
        data = result.get("data", [{}])[0]
        malicious = data.get("summary", {}).get("malicious", 0)

        if malicious > 3:
            phantom.act("block ip", targets=result,
                        assets=["palo_alto_prod"],
                        name="block_malicious_ip")
```

### Step 3: Build Alert Enrichment Playbook

Automate enrichment for all incoming SIEM alerts:

```python
"""
Universal Alert Enrichment Playbook
Runs on every new event to add context before analyst review
"""

import phantom.rules as phantom

def on_start(container):
    # Get all artifacts
    success, message, artifacts = phantom.get_artifacts(
        container_id=container["id"], full_data=True
    )

    ip_artifacts = [a for a in artifacts if a.get("cef", {}).get("sourceAddress")]
    domain_artifacts = [a for a in artifacts if a.get("cef", {}).get("destinationDnsDomain")]

    # Enrich IPs in parallel
    for artifact in ip_artifacts:
        ip = artifact["cef"]["sourceAddress"]

        # VirusTotal lookup
        phantom.act("ip reputation",
                    parameters=[{"ip": ip}],
                    assets=["virustotal_prod"],
                    callback=enrich_ip_callback,
                    name=f"vt_ip_{ip}")

        # GeoIP lookup
        phantom.act("geolocate ip",
                    parameters=[{"ip": ip}],
                    assets=["maxmind_prod"],
                    callback=geoip_callback,
                    name=f"geo_{ip}")

        # Whois lookup
        phantom.act("whois ip",
                    parameters=[{"ip": ip}],
                    assets=["whois_prod"],
                    name=f"whois_{ip}")

    # Enrich domains
    for artifact in domain_artifacts:
        domain = artifact["cef"]["destinationDnsDomain"]
        phantom.act("domain reputation",
                    parameters=[{"domain": domain}],
                    assets=["virustotal_prod"],
                    name=f"vt_domain_{domain}")

def enrich_ip_callback(action, success, container, results, handle):
    """Update container with enrichment data"""
    if success:
        for result in results:
            summary = result.get("summary", {})
            phantom.add_artifact(container, {
                "cef": {
                    "vt_malicious": summary.get("malicious", 0),
                    "vt_suspicious": summary.get("suspicious", 0),
                    "enrichment_source": "VirusTotal"
                },
                "label": "enrichment",
                "name": "VT IP Enrichment"
            })
```

### Step 4: Implement Approval Gates for High-Impact Actions

Add human-in-the-loop for critical actions:

```python
def containment_decision(action, success, container, results, handle):
    """Present analyst with containment options"""
    phantom.prompt(
        container=container,
        user="soc_tier2",
        message=(
            "Confirmed malicious activity detected.\n"
            f"Host: {container['artifacts'][0]['cef'].get('sourceAddress')}\n"
            f"Threat: {results[0]['summary'].get('threat_name')}\n\n"
            "Select containment action:"
        ),
        respond_in_mins=15,
        options=["Isolate Host", "Disable Account", "Both", "Monitor Only"],
        callback=execute_containment
    )

def execute_containment(action, success, container, results, handle):
    response = results.get("response", "Monitor Only")

    if response in ["Isolate Host", "Both"]:
        phantom.act("quarantine device",
                    parameters=[{"hostname": container["artifacts"][0]["cef"]["sourceHostName"]}],
                    assets=["crowdstrike_prod"],
                    name="isolate_host")

    if response in ["Disable Account", "Both"]:
        phantom.act("disable user",
                    parameters=[{"username": container["artifacts"][0]["cef"]["sourceUserName"]}],
                    assets=["ad_prod"],
                    name="disable_account")

    phantom.comment(container, f"Analyst approved: {response}")
```

### Step 5: Configure Playbook Scheduling and Triggers

Set up event triggers in SOAR:

```json
{
  "playbook_name": "phishing_triage_automation",
  "trigger": {
    "type": "event_created",
    "conditions": {
      "label": ["phishing", "notable"],
      "severity": ["high", "medium"]
    }
  },
  "active": true,
  "run_as": "automation_user"
}
```

### Step 6: Monitor Playbook Performance

Track automation effectiveness with SOAR metrics:

```python
# Query SOAR API for playbook execution stats
import requests

headers = {"ph-auth-token": "YOUR_SOAR_TOKEN"}
response = requests.get(
    "https://soar.company.com/rest/playbook_run",
    headers=headers,
    params={
        "page_size": 100,
        "filter": '{"status":"success"}',
        "sort": "create_time",
        "order": "desc"
    }
)
runs = response.json()["data"]

# Calculate automation metrics
total_runs = len(runs)
avg_duration = sum(r["end_time"] - r["start_time"] for r in runs) / total_runs
auto_closed = sum(1 for r in runs if r.get("auto_resolved"))
print(f"Total runs: {total_runs}")
print(f"Avg duration: {avg_duration:.1f}s")
print(f"Auto-resolved: {auto_closed}/{total_runs} ({auto_closed/total_runs*100:.0f}%)")
```

## Key Concepts

| Term | Definition |
|------|-----------|
| **SOAR** | Security Orchestration, Automation, and Response — platform integrating security tools with automated playbooks |
| **Playbook** | Automated workflow defining sequential and parallel actions triggered by security events |
| **Asset** | SOAR configuration for a connected security tool (API endpoint, credentials, connection parameters) |
| **Container** | SOAR event object containing artifacts (IOCs) from an ingested alert or incident |
| **Artifact** | Individual IOC or data point within a container (IP, hash, URL, domain, email) |
| **Approval Gate** | Human-in-the-loop step requiring analyst decision before executing high-impact automated actions |

## Tools & Systems

- **Splunk SOAR (Phantom)**: Enterprise SOAR platform with 300+ app integrations and visual playbook editor
- **Splunk ES**: SIEM platform feeding notable events into SOAR as containers for automated triage
- **CrowdStrike Falcon**: EDR platform integrated via SOAR for automated host isolation and threat hunting
- **ServiceNow**: ITSM platform integrated for automated incident ticket creation and tracking
- **Palo Alto NGFW**: Firewall integrated for automated IP/URL blocking via SOAR playbooks

## Common Scenarios

- **Phishing Triage**: Auto-extract URLs/attachments, detonate in sandbox, block malicious, create ticket
- **Malware Alert Enrichment**: Auto-enrich file hashes across VT/MalwareBazaar, isolate if confirmed malicious
- **Brute Force Response**: Auto-check if attack succeeded, disable account if compromised, block source IP
- **Threat Intel IOC Processing**: Auto-ingest TI feed IOCs, check against internal logs, create blocks for matches
- **Vulnerability Alert Response**: Auto-query asset database for affected systems, create patching ticket with priority

## Output Format

```
SOAR PLAYBOOK EXECUTION REPORT
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Playbook:     Phishing Triage Automation v2.3
Container:    SOAR-2024-08921
Trigger:      Notable event from Splunk ES (phishing)

Actions Executed:
  [1] URL Reputation (VirusTotal)     — 14/90 engines malicious    [2.1s]
  [2] IP Reputation (AbuseIPDB)       — Confidence: 85%            [1.3s]
  [3] Block URL (Palo Alto)           — Blocked on PA-5260         [0.8s]
  [4] Block IP (Palo Alto)            — Blocked on PA-5260         [0.7s]
  [5] Create Ticket (ServiceNow)      — INC0012345 created         [1.5s]
  [6] Prompt Analyst (Tier 2)         — Response: "Isolate Host"   [4m 12s]
  [7] Quarantine Device (CrowdStrike) — WORKSTATION-042 isolated   [3.2s]

Total Duration:    4m 22s (vs 35min avg manual triage)
Time Saved:        ~31 minutes
Disposition:       True Positive — Escalated to IR
```