Malware Tracker automates deep stream dissection across Microsoft Office (OLE2, OOXML, RTF, XLM macros, encrypted VelvetSweatshop containers), PDF filter chains, and MIME emails—combining the heritage of Cryptam and PDFExaminer into a single in-memory engine and REST API.
Originally built by @tylabs to dissect targeted APT spear-phishing attachments, Malware Tracker's static cryptanalysis and PDF object parsers now run as a unified, memory-safe Python framework and high-throughput Cloud Run API.
Unpacks OLE2, OpenXML, RTF, CHM, HLP, VBA stomping, Excel 4.0 XLM macros, and encrypted Office containers. Detects embedded executables obscured via XOR, bitwise ROL/ROR shifts, NOT, and header transposition ciphers.
Steps through obfuscated PDF object trees, recursively decodes compressed filter chains (/FlateDecode, /ASCII85, PNG predictors), de-obfuscates JavaScript, and flags invisible 3 Tr AI prompt-injection streams.
Combines deep recursive container unpacking with 239 YARA exploit, shellcode, and AI/RAG safety rules. Runs in <50ms in memory via CLI, Python library, or our zero-retention REST API (POST /v1/scan).
Embed QuickSand directly into web application upload pipelines, S3/GCS cloud storage buckets, SOAR playbooks, and LLM RAG ingestion guardrails.
/v1/scan)Inspect user-submitted resumes, invoices, tax forms, and loan documents in <100ms via REST API before writing to disk. Default ?store=false zero-retention execution satisfies strict SOC 2 and HIPAA controls.
Replace 4 GB ClamAV containers with a lightweight serverless function triggered on object creation. Scans incoming files in ~40ms and routes clean documents or isolates weaponized uploads into quarantine buckets.
Terraform Templates →Run the open-source MIT quicksand engine locally and pull our commercial YARA signature feed (GET /v1/rules/pro) covering 67+ Office/PDF CVEs (CVE-2025-24071, CVE-2025-8088) and shellcode hashes.
/v1/triage)Pre-filter 100% of phishing email attachments in Tines, Shuffle, n8n, or Cortex XSOAR in milliseconds. Returns a deterministic 0–3 risk rating so you only detonate genuinely suspicious files in expensive VMs.
Detects hidden LLM instruction overrides inside PDFs and Word documents—including PDF invisible rendering mode (3 Tr), micro/zero-point fonts, OOXML <w:vanish/> text, and Markdown data exfiltration links.
Traditional signature AV fails on encrypted Office lures, nested PDF filter chains, and VBA stomping while consuming 2–4 GB RAM. Read our stream-level detection and latency comparison.
Read Benchmark →Single-request HTTP inspection with structured JSON verdicts, MITRE ATT&CK mappings, and stream offsets.
# Synchronous in-memory scan with zero file retention (?store=false)
curl -X POST "https://api.malwaretracker.com/v1/scan?store=false" \
-H "X-API-Key: mt_live_YOUR_API_KEY" \
-F "file=@attachment.pdf"
# Response (38ms):
{
"verdict": "malicious",
"rating": 3,
"score": 29,
"scan_time_ms": 38.4,
"zero_retention": true,
"threat_intel": {
"cves": ["CVE-2025-8088"],
"mitre_attack": ["T1059", "T1203"],
"streams_analyzed": 14
},
"ai_safety": {
"safe": false,
"prompt_injection_detected": true,
"findings": [
{
"rule": "ai_pdf_invisible_or_zero_font_text",
"description": "AI/RAG evasion: PDF stream uses invisible text rendering mode (3 Tr)"
}
]
}
}
// Next.js / Express Upload Verification
export async function verifyUploadedDocument(fileBuffer, filename) {
const form = new FormData();
form.append("file", new Blob([fileBuffer]), filename);
const res = await fetch(
"https://api.malwaretracker.com/v1/scan?store=false",
{
method: "POST",
headers: { "X-API-Key": process.env.MALWARETRACKER_API_KEY },
body: form,
}
);
const report = await res.json();
if (report.verdict !== "clean" || !report.ai_safety.safe) {
throw new Error(`Upload rejected: ${report.verdict} (${report.threat_intel.cves.join(", ")})`);
}
return report;
}
import requests
def scan_document_before_ingestion(file_path: str, api_key: str) -> dict:
with open(file_path, "rb") as f:
resp = requests.post(
"https://api.malwaretracker.com/v1/scan",
headers={"X-API-Key": api_key},
params={"store": "false"},
files={"file": f},
timeout=15,
)
resp.raise_for_status()
data = resp.json()
if data["verdict"] != "clean" or not data["ai_safety"]["safe"]:
raise PermissionError(f"Blocked untrusted document: {data['verdict']}")
return data
# Tier-0 Static Triage for Tines, Shuffle, n8n, or Cortex XSOAR
curl -X POST "https://api.malwaretracker.com/v1/triage" \
-H "X-API-Key: mt_live_YOUR_API_KEY" \
-F "file=@reported_phish.doc"
# Routing Verdict:
{
"verdict": "malicious",
"rating": 3,
"escalate_to_sandbox": true,
"recommended_action": "quarantine_and_detonate",
"ai_prompt_injection": false,
"cves": ["CVE-2025-24071"],
"mitre_attack": ["T1187", "T1203"],
"scan_time_ms": 24.1
}
# Download commercial QuickSand Pro Zero-Day CVE & AI Safety YARA bundle curl -sSL "https://api.malwaretracker.com/v1/rules/pro" \ -H "X-API-Key: mt_live_YOUR_PRO_KEY" \ -o quicksand_pro_rules.yara # Run locally with open-source QuickSand CLI: quicksand suspicious.doc --exploityara quicksand_pro_rules.yara
Instant self-serve provisioning backed by Stripe. No enterprise sales calls required.
For security researchers, evaluation, and low-volume apps.
POST /v1/scan (<100ms API)?store=false modeFor applications scanning user-uploaded PDFs and Office docs.
/v1/triage)High-volume API + QuickSand Pro YARA feed + Bucket Guard.
/v1/rules/pro)Annual commercial YARA signature feed for internal pipelines.
/v1/rules/pro)