The Agent Skills Directory

[DATA_EXFILTRATION]: Provides explicit instructions and XML payloads for reading sensitive system files including /etc/passwd, /etc/shadow, and Windows configuration files (win.ini).
[DATA_EXFILTRATION]: Describes methods for exfiltrating sensitive data to external, non-whitelisted domains (e.g., YOUR.BURPCOLLABORATOR.net, attacker.com) using Out-of-Band (OOB) HTTP and DNS callbacks.
[DATA_EXFILTRATION]: Targets cloud-specific sensitive data, providing payloads to access AWS instance metadata and IAM security credentials at 169.254.169.254.
[COMMAND_EXECUTION]: Includes curl command examples designed to transmit XML payloads to remote endpoints for vulnerability testing.
[OBFUSCATION]: Details techniques for bypassing security filters using hex-encoded entities (e.g., file for file) and variations in character encoding (UTF-16).
[DYNAMIC_EXECUTION]: Mentions the use of vulnerable PHP and Java functions (e.g., simplexml_load_string, rt:exec) as part of XXE-to-RCE exploitation chains.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and process data from untrusted external sources, creating an attack surface for indirect prompt injection.
Ingestion points: XML entry points, file upload features (SVG, DOCX, XLSX), and SAML/SSO endpoints as described throughout SKILL.md.
Boundary markers: Absent. The instructions do not provide delimiters or warnings to treat external target responses as non-instructional data.
Capability inventory: File system read access and network connectivity via curl and XML external entity resolution.
Sanitization: Absent. No sanitization or validation protocols are defined for handling content retrieved from external targets.

hunt-xxe