npx skills add mukul975/Anthropic-Cybersecurity-SkillsMITRE ATT&CK
NIST CSF 2.0
MITRE D3FEND
When to Use
- Interactive malware analysis is needed where the analyst must click dialogs, enter credentials, or navigate installer screens
- Rapid cloud-based sandbox analysis without maintaining local sandbox infrastructure
- Malware requires user interaction to proceed past anti-sandbox checks (document macros requiring "Enable Content")
- Sharing analysis results with team members via public or private task URLs
- Comparing behavior across different OS versions (Windows 7, 10, 11) available in ANY.RUN
Do not use for highly sensitive samples that cannot be uploaded to cloud services; use an on-premises sandbox like Cuckoo instead.
Prerequisites
- ANY.RUN account (free community tier or paid subscription at https://any.run)
- Modern web browser with WebSocket support for interactive session streaming
- Sample file ready for upload (max 100 MB for free tier, 256 MB for paid)
- Understanding of the sample type to select appropriate execution environment
- VPN or secure network for accessing ANY.RUN portal during analysis sessions
Workflow
Step 1: Configure Analysis Environment
Set up the ANY.RUN task with appropriate parameters:
ANY.RUN Task Configuration:
━━━━━━━━━━━━━━━━━━━━━━━━━━
OS Selection: Windows 10 x64 (recommended default)
Windows 7 x64 (for legacy malware)
Windows 11 x64 (for modern samples)
Execution Time: 60 seconds (default) / 120-300 for slow-acting malware
Network: Connected (captures real C2 traffic)
Residential Proxy (bypasses geo-blocking)
Privacy: Public (free tier) / Private (paid - not indexed)
MITM Proxy: Enable for HTTPS traffic decryption
Fake Net: Enable to simulate internet services if sample checks connectivityAPI-based submission (paid tier):
# Submit file via ANY.RUN API
curl -X POST "https://api.any.run/v1/analysis" \
-H "Authorization: API-Key $ANYRUN_API_KEY" \
-F "file=@suspect.exe" \
-F "env_os=windows" \
-F "env_version=10" \
-F "env_bitness=64" \
-F "opt_timeout=120" \
-F "opt_network_connect=true" \
-F "opt_privacy_type=bylink"
# Check task status
curl "https://api.any.run/v1/analysis/$TASK_ID" \
-H "Authorization: API-Key $ANYRUN_API_KEY" | jq '.data.status'Step 2: Interact with Malware During Execution
Use the interactive session to trigger malware behavior:
Interactive Actions During Analysis:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
1. Document Macros: Click "Enable Content" / "Enable Editing" when prompted
2. Installer Screens: Click through installation dialogs
3. UAC Prompts: Click "Yes" to allow elevation (observe privilege escalation)
4. Credential Harvests: Enter fake credentials to observe phishing behavior
5. Browser Redirects: Navigate to URLs if malware opens browser windows
6. File Dialogs: Select target files if malware presents file picker
7. Timeout Extension: Extend analysis time if malware has delayed executionStep 3: Analyze Process Tree
Review the complete process execution chain:
Process Tree Analysis Points:
━━━━━━━━━━━━━━━━━━━━━━━━━━━
Parent-Child Relationships:
- WINWORD.EXE -> cmd.exe -> powershell.exe (macro execution chain)
- explorer.exe -> suspect.exe -> svchost.exe (process injection)
Process Events to Note:
- Process creation with suspicious command-line arguments
- PowerShell with encoded commands (-enc / -encodedcommand)
- cmd.exe executing script files (.bat, .vbs, .js)
- Legitimate processes spawned from unusual parents
- Process termination (self-deletion behavior)Step 4: Review Network Activity
Examine DNS, HTTP/HTTPS, and TCP/UDP connections:
ANY.RUN Network Panel Analysis:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
DNS Requests:
- Domain resolutions with threat intelligence tags
- Fast-flux or DGA domain patterns
- DNS over HTTPS (DoH) detection
HTTP/HTTPS Traffic (with MITM enabled):
- Full request/response bodies for HTTP
- Decrypted HTTPS traffic showing C2 commands
- Downloaded payloads and their content types
- POST data containing exfiltrated information
Connection Map:
- Geographic visualization of C2 server locations
- Connection timeline showing beacon patterns
- Suricata alerts triggered on network trafficStep 5: Examine IOCs and Threat Intelligence
Extract indicators and map to known threats:
ANY.RUN IOC Categories:
━━━━━━━━━━━━━━━━━━━━━━
Files: Dropped files with hashes, YARA matches, VirusTotal results
Network: IPs, domains, URLs contacted during execution
Registry: Keys created/modified for persistence
Processes: Suspicious process names and command lines
Mutex: Named mutexes created (used for single-instance checking)
Signatures: Suricata rules triggered, behavioral signatures matched
MITRE ATT&CK Mapping:
- ANY.RUN automatically maps observed behaviors to ATT&CK techniques
- Review the ATT&CK matrix tab for technique coverage
- Export ATT&CK Navigator layer for reportingStep 6: Export Analysis Results
Download comprehensive reports and artifacts:
# Download report via API
curl "https://api.any.run/v1/analysis/$TASK_ID/report" \
-H "Authorization: API-Key $ANYRUN_API_KEY" \
-o report.json
# Download PCAP
curl "https://api.any.run/v1/analysis/$TASK_ID/pcap" \
-H "Authorization: API-Key $ANYRUN_API_KEY" \
-o capture.pcap
# Download dropped files
curl "https://api.any.run/v1/analysis/$TASK_ID/files" \
-H "Authorization: API-Key $ANYRUN_API_KEY" \
-o dropped_files.zip
# Available exports from ANY.RUN web interface:
# - HTML Report (shareable standalone page)
# - PCAP file (network traffic capture)
# - Process dump (memory dumps of processes)
# - Dropped files (all files created during execution)
# - MITRE ATT&CK Navigator JSON
# - IOC export (STIX/JSON/CSV format)Key Concepts
| Term | Definition |
|---|---|
| Interactive Sandbox | Analysis environment allowing real-time analyst interaction with the executing sample, enabling triggering of user-dependent behaviors |
| MITM Proxy | Man-in-the-middle TLS interception in ANY.RUN that decrypts HTTPS traffic for visibility into encrypted C2 communications |
| Residential Proxy | ANY.RUN feature routing malware traffic through residential IP addresses to bypass geo-IP and datacenter-IP evasion checks |
| Suricata Alerts | Network IDS signatures triggered during execution, providing immediate identification of known malicious traffic patterns |
| Process Tree | Hierarchical visualization of parent-child process relationships showing the complete execution chain from initial sample to final payloads |
| Behavioral Tags | ANY.RUN classification labels automatically applied based on observed behavior (e.g., "trojan", "stealer", "ransomware") |
Tools & Systems
- ANY.RUN: Cloud-based interactive malware sandbox providing real-time execution monitoring, process trees, network capture, and MITRE ATT&CK mapping
- ANY.RUN API: REST API for programmatic sample submission, status checking, and report/artifact retrieval
- Suricata: Integrated network IDS within ANY.RUN providing signature-based detection of malicious network traffic
- MITRE ATT&CK Navigator: Framework integration mapping observed malware behaviors to adversary techniques and tactics
- VirusTotal Integration: Automatic hash lookup of sample and dropped files against VirusTotal detection results
Common Scenarios
Scenario: Analyzing a Macro-Enabled Document Requiring User Interaction
Context: Phishing email contains a .docm file that requires clicking "Enable Content" to trigger the macro payload. Traditional non-interactive sandboxes fail to trigger the malicious behavior.
Approach:
- Upload .docm to ANY.RUN with Windows 10 environment and Microsoft Office installed
- When Word opens and displays the security banner, click "Enable Content" interactively
- Observe the macro execution in the process tree (Word -> cmd.exe -> powershell.exe)
- Monitor network panel for PowerShell downloading second-stage payload
- If a UAC prompt appears, click "Yes" to allow the payload to observe full behavior chain
- Review Suricata alerts for known malware signatures on the downloaded payload
- Export IOCs (download URLs, dropped file hashes, C2 domains) for blocking
Pitfalls:
- Forgetting to enable MITM proxy, resulting in encrypted HTTPS traffic without visibility
- Using too short an execution timeout for malware with delayed execution or sleep timers
- Uploading to public analysis when the sample contains sensitive organizational data
- Not clicking through all prompts; some malware requires multiple user interactions to fully execute
Output Format
ANY.RUN ANALYSIS REPORT
=========================
Task URL: https://app.any.run/tasks/<task_id>
Sample: invoice_q3.docm
SHA-256: e3b0c44298fc1c149afbf4c8996fb924...
Verdict: MALICIOUS (Score: 95/100)
Family: Emotet
Tags: [trojan, banker, spam, macro]
PROCESS TREE
WINWORD.EXE (PID: 2184)
└── cmd.exe (PID: 3456) "/c powershell -enc JABXAG..."
└── powershell.exe (PID: 4012)
└── rundll32.exe (PID: 4568) "C:\Users\...\payload.dll,Control_RunDLL"
NETWORK INDICATORS
DNS: update.emotet-c2[.]com -> 185.220.101.42
HTTPS: POST hxxps://185.220.101[.]42/wp-content/gate/ (C2 beacon)
HTTP: GET hxxp://compromised-site[.]com/invoice.dll (payload download)
SURICATA ALERTS
[1:2028401] ET MALWARE Emotet CnC Beacon
[1:2028402] ET MALWARE Win32/Emotet Activity
MITRE ATT&CK TECHNIQUES
T1566.001 Phishing: Spearphishing Attachment
T1204.002 User Execution: Malicious File
T1059.001 Command and Scripting Interpreter: PowerShell
T1218.011 Rundll32 Execution
T1071.001 Application Layer Protocol: Web Protocols
DROPPED FILES
payload.dll SHA-256: abc123... Detection: 48/72 (VirusTotal)
config.dat SHA-256: def456... (encrypted configuration)References and resources
Everything below is rendered for inspection. Script files are read-only and never run.
References 1
api-reference.md3.1 KB
API Reference: Performing Dynamic Analysis with ANY.RUN
ANY.RUN API v1
| Endpoint | Method | Description |
|---|---|---|
/v1/analysis |
POST | Submit file or URL for analysis |
/v1/analysis/{taskid} |
GET | Get full analysis report |
/v1/analysis/{taskid}/ioc |
GET | Get extracted IOCs |
/v1/analysis/{taskid}/download/{type} |
GET | Download PCAP, screenshots, or dropped files |
Submission Parameters
| Parameter | Type | Description |
|---|---|---|
file |
file | Malware sample to analyze (multipart upload) |
obj_url |
string | URL to analyze in browser |
env_os |
string | OS: windows-7, windows-10, windows-11 |
env_bitness |
int | Architecture: 32 or 64 |
opt_privacy_type |
string | public, private, or bylink |
opt_timeout |
int | Analysis timeout in seconds (60-660) |
opt_network_connect |
bool | Allow internet access during analysis |
opt_network_fakenet |
bool | Use fake network services |
Report Structure
| Field | Description |
|---|---|
analysis.scores.verdict |
Overall verdict and threat level |
analysis.processes[] |
Process tree with command lines |
analysis.network.dnsRequests[] |
DNS queries made by sample |
analysis.network.httpRequests[] |
HTTP requests with URLs and methods |
analysis.network.connections[] |
TCP/UDP connections |
analysis.mitre[] |
Mapped MITRE ATT&CK techniques |
analysis.tags[] |
Malware family and behavior tags |
Official Python SDK (anyrun-sdk)
| Class / Method | Description |
|---|---|
SandboxConnector.windows(api_key) |
Create sandbox connector for Windows analysis (context manager) |
SandboxConnector.linux(api_key) |
Create sandbox connector for Linux analysis (context manager) |
connector.run_file_analysis(filepath) |
Submit local file, returns analysis_id |
connector.run_url_analysis(url) |
Submit URL for browser analysis, returns analysis_id |
connector.get_task_status(analysis_id) |
Generator yielding status updates until completion |
connector.get_analysis_verdict(analysis_id) |
Returns verdict string (malicious/suspicious/clean) |
connector.get_analysis_report(analysis_id) |
Returns full analysis report dict |
Key Libraries
- anyrun-sdk (
pip install anyrun-sdk): Official ANY.RUN Python SDK withSandboxConnector - requests (
pip install requests): HTTP client for REST API fallback - time (stdlib): Polling for analysis completion
- json (stdlib): Parse and export analysis results
Configuration
| Variable | Description |
|---|---|
ANYRUN_API_KEY |
ANY.RUN API key (from account settings) |
Rate Limits
| Plan | Submissions/Day | API Calls/Minute |
|---|---|---|
| Free | 5 public | 10 |
| Hunter | Unlimited private | 60 |
| Enterprise | Unlimited | 120 |
References
Scripts 1
agent.py10.5 KB
#!/usr/bin/env python3
"""
Dynamic Analysis with ANY.RUN Agent
Submits malware samples to ANY.RUN sandbox via API, monitors task execution,
and retrieves behavioral analysis results including process trees, network
indicators, and MITRE ATT&CK mappings.
Supports both the official anyrun-sdk (pip install anyrun-sdk) with
SandboxConnector and the legacy REST API via requests.
"""
import json
import os
import sys
import time
from datetime import datetime, timezone
import requests
try:
from anyrun.connectors import SandboxConnector
HAS_ANYRUN_SDK = True
except ImportError:
HAS_ANYRUN_SDK = False
ANYRUN_API_BASE = "https://api.any.run/v1"
def submit_file(filepath: str, api_key: str, os_version: str = "windows-10",
privacy: str = "private", timeout_seconds: int = 120) -> dict:
"""Submit a file to ANY.RUN for dynamic analysis."""
if not os.path.exists(filepath):
return {"error": f"File not found: {filepath}"}
headers = {"Authorization": f"API-Key {api_key}"}
with open(filepath, "rb") as f:
files = {"file": (os.path.basename(filepath), f)}
data = {
"env_os": os_version,
"env_bitness": 64,
"env_type": "complete",
"opt_privacy_type": privacy,
"opt_timeout": timeout_seconds,
"opt_network_connect": True,
"opt_network_fakenet": False,
"opt_network_tor": False,
}
resp = requests.post(
f"{ANYRUN_API_BASE}/analysis",
headers=headers, files=files, data=data, timeout=60,
)
if resp.status_code in (200, 201):
result = resp.json()
return {
"task_id": result.get("data", {}).get("taskid", ""),
"status": "submitted",
"task_url": f"https://app.any.run/tasks/{result.get('data', {}).get('taskid', '')}",
}
return {"error": f"Submission failed: {resp.status_code} - {resp.text[:200]}"}
def submit_url(url: str, api_key: str, os_version: str = "windows-10") -> dict:
"""Submit a URL to ANY.RUN for dynamic analysis."""
headers = {"Authorization": f"API-Key {api_key}"}
data = {
"obj_url": url,
"env_os": os_version,
"env_bitness": 64,
"opt_privacy_type": "private",
"opt_timeout": 120,
"opt_network_connect": True,
}
resp = requests.post(
f"{ANYRUN_API_BASE}/analysis",
headers=headers, data=data, timeout=60,
)
if resp.status_code in (200, 201):
result = resp.json()
return {
"task_id": result.get("data", {}).get("taskid", ""),
"status": "submitted",
}
return {"error": f"URL submission failed: {resp.status_code}"}
def get_task_report(task_id: str, api_key: str) -> dict:
"""Retrieve the full analysis report for a completed task."""
headers = {"Authorization": f"API-Key {api_key}"}
resp = requests.get(
f"{ANYRUN_API_BASE}/analysis/{task_id}",
headers=headers, timeout=30,
)
if resp.status_code != 200:
return {"error": f"Report retrieval failed: {resp.status_code}"}
data = resp.json().get("data", {})
report = {
"task_id": task_id,
"verdict": data.get("analysis", {}).get("scores", {}).get("verdict", {}).get("verdict", "unknown"),
"threat_level": data.get("analysis", {}).get("scores", {}).get("verdict", {}).get("threatLevelText", ""),
"tags": data.get("analysis", {}).get("tags", []),
}
processes = data.get("analysis", {}).get("processes", [])
report["processes"] = []
for proc in processes:
report["processes"].append({
"pid": proc.get("pid", 0),
"name": proc.get("fileName", ""),
"command_line": proc.get("commandLine", "")[:200],
"parent_pid": proc.get("parentPID", 0),
"is_malicious": proc.get("scores", {}).get("verdict", {}).get("isMalicious", False),
})
network = data.get("analysis", {}).get("network", {})
report["network"] = {
"dns_requests": [
{"domain": d.get("domain", ""), "ip": d.get("ip", "")}
for d in network.get("dnsRequests", [])
],
"http_requests": [
{"url": h.get("url", ""), "method": h.get("method", ""), "status": h.get("status", 0)}
for h in network.get("httpRequests", [])
],
"connections": [
{"ip": c.get("ip", ""), "port": c.get("port", 0), "protocol": c.get("protocol", "")}
for c in network.get("connections", [])
],
}
mitre = data.get("analysis", {}).get("mitre", [])
report["mitre_techniques"] = [
{"technique_id": m.get("id", ""), "name": m.get("name", ""), "tactic": m.get("tactic", "")}
for m in mitre
]
return report
def wait_for_completion(task_id: str, api_key: str, max_wait: int = 300) -> dict:
"""Poll task status until analysis completes."""
headers = {"Authorization": f"API-Key {api_key}"}
start = time.time()
while time.time() - start < max_wait:
resp = requests.get(
f"{ANYRUN_API_BASE}/analysis/{task_id}",
headers=headers, timeout=30,
)
if resp.status_code == 200:
status = resp.json().get("data", {}).get("analysis", {}).get("status", "")
if status == "done":
return {"status": "completed", "elapsed": round(time.time() - start)}
if status == "failed":
return {"status": "failed", "elapsed": round(time.time() - start)}
time.sleep(15)
return {"status": "timeout", "elapsed": max_wait}
def get_iocs_from_report(report: dict) -> dict:
"""Extract IOCs from an ANY.RUN analysis report."""
iocs = {
"domains": set(),
"ips": set(),
"urls": set(),
"processes": [],
"mitre_techniques": [],
}
for dns_req in report.get("network", {}).get("dns_requests", []):
if dns_req.get("domain"):
iocs["domains"].add(dns_req["domain"])
if dns_req.get("ip"):
iocs["ips"].add(dns_req["ip"])
for http_req in report.get("network", {}).get("http_requests", []):
if http_req.get("url"):
iocs["urls"].add(http_req["url"])
for conn in report.get("network", {}).get("connections", []):
if conn.get("ip"):
iocs["ips"].add(conn["ip"])
for proc in report.get("processes", []):
if proc.get("is_malicious"):
iocs["processes"].append(proc["name"])
iocs["mitre_techniques"] = report.get("mitre_techniques", [])
iocs["domains"] = sorted(iocs["domains"])
iocs["ips"] = sorted(iocs["ips"])
iocs["urls"] = sorted(iocs["urls"])
return iocs
def generate_report(submission: dict, report: dict, iocs: dict) -> str:
"""Generate dynamic analysis report."""
lines = [
"DYNAMIC ANALYSIS REPORT (ANY.RUN)",
"=" * 50,
f"Date: {datetime.now(timezone.utc).strftime('%Y-%m-%d %H:%M UTC')}",
f"Task ID: {report.get('task_id', submission.get('task_id', 'N/A'))}",
f"Task URL: {submission.get('task_url', 'N/A')}",
"",
f"Verdict: {report.get('verdict', 'N/A')}",
f"Threat Level: {report.get('threat_level', 'N/A')}",
f"Tags: {', '.join(report.get('tags', []))}",
"",
f"PROCESSES ({len(report.get('processes', []))}):",
]
for proc in report.get("processes", [])[:10]:
mal = " [MALICIOUS]" if proc.get("is_malicious") else ""
lines.append(f" PID {proc['pid']}: {proc['name']}{mal}")
if proc.get("command_line"):
lines.append(f" CMD: {proc['command_line'][:100]}")
lines.extend([
"",
"NETWORK IOCs:",
f" Domains: {len(iocs.get('domains', []))}",
f" IPs: {len(iocs.get('ips', []))}",
f" URLs: {len(iocs.get('urls', []))}",
])
if iocs.get("mitre_techniques"):
lines.extend(["", "MITRE ATT&CK TECHNIQUES:"])
for tech in iocs["mitre_techniques"][:10]:
lines.append(f" {tech['technique_id']} - {tech['name']} ({tech['tactic']})")
return "\n".join(lines)
def run_with_sdk(api_key: str, target: str, is_url: bool) -> dict:
"""Use official anyrun-sdk (SandboxConnector) if available."""
with SandboxConnector.windows(api_key) as connector:
if is_url:
analysis_id = connector.run_url_analysis(target)
else:
analysis_id = connector.run_file_analysis(target)
print(f"[*] SDK analysis ID: {analysis_id}")
for status in connector.get_task_status(analysis_id):
print(f"[*] Status: {status}")
verdict = connector.get_analysis_verdict(analysis_id)
report = connector.get_analysis_report(analysis_id)
return {"analysis_id": analysis_id, "verdict": verdict, "report": report}
if __name__ == "__main__":
api_key = os.getenv("ANYRUN_API_KEY", "")
if not api_key:
print("[!] Set ANYRUN_API_KEY environment variable")
sys.exit(1)
if len(sys.argv) < 2:
print(f"Usage: {sys.argv[0]} <file_or_url> [--url]")
sys.exit(1)
target = sys.argv[1]
is_url = "--url" in sys.argv or target.startswith("http")
# Prefer official SDK when available, fall back to REST API
if HAS_ANYRUN_SDK:
print("[*] Using official anyrun-sdk (SandboxConnector)")
sdk_result = run_with_sdk(api_key, target, is_url)
print(json.dumps(sdk_result, indent=2, default=str))
sys.exit(0)
print("[*] anyrun-sdk not found, using REST API fallback")
if is_url:
print(f"[*] Submitting URL: {target}")
submission = submit_url(target, api_key)
else:
print(f"[*] Submitting file: {target}")
submission = submit_file(target, api_key)
if "error" in submission:
print(f"[!] {submission['error']}")
sys.exit(1)
task_id = submission["task_id"]
print(f"[*] Task ID: {task_id}")
print("[*] Waiting for analysis to complete...")
completion = wait_for_completion(task_id, api_key)
print(f"[*] Status: {completion['status']} ({completion.get('elapsed', 0)}s)")
if completion["status"] == "completed":
report = get_task_report(task_id, api_key)
iocs = get_iocs_from_report(report)
output_text = generate_report(submission, report, iocs)
print(output_text)
output = f"anyrun_analysis_{task_id}.json"
with open(output, "w") as f:
json.dump({"submission": submission, "report": report, "iocs": iocs}, f, indent=2)
print(f"\n[*] Results saved to {output}")
else:
print(f"[!] Analysis did not complete: {completion['status']}")