hunt-xxe
Warn
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONCOMMAND_EXECUTIONOBFUSCATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [DATA_EXFILTRATION]: Provides offensive payloads designed to exfiltrate sensitive system files such as /etc/passwd and C:/Windows/win.ini.
- Includes payloads for Out-of-Band (OOB) exfiltration that transmit local file content to attacker-controlled listener services (e.g., Burp Collaborator).
- Targets cloud environment metadata via the 169.254.169.254 endpoint to harvest instance credentials.
- [COMMAND_EXECUTION]: Recommends the use of shell commands such as curl for exploiting endpoints and grep for scanning local source code for vulnerable XML parser configurations.
- [OBFUSCATION]: Details various techniques to evade security filters and Web Application Firewalls (WAFs).
- Includes the use of character encoding (UTF-16) and nested hex-encoded entities (e.g., file) to hide malicious payloads.
- Suggests using chunked transfer encoding and malformed XML comments to break signature-based detection.
- [INDIRECT_PROMPT_INJECTION]: This skill is designed to process untrusted XML data, creating an attack surface for indirect prompt injection.
- Ingestion points: Processes XML content types, RSS/Sitemap feeds, and various file formats (SVG, DOCX, XLSX) that act as XML containers.
- Boundary markers: Implements a mandatory confirmation gate requiring the user to verify target scope and authorization before execution.
- Capability inventory: Capable of performing file reads, SSRF requests, and network callbacks to external infrastructure.
- Sanitization: While focusing on exploitation, it mentions defensive hardening practices such as disabling external entities.
- [DYNAMIC_EXECUTION]: Discusses advanced exploitation involving XSLT extension execution and nested deserialization (e.g., CVE-2024-34102) to achieve remote code execution.
Audit Metadata