xslt-injection
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a comprehensive technical guide for security professionals to test for XSLT injection vulnerabilities. While it includes functional code snippets for file reading, command execution, and network requests, these are presented as explicit templates for testing in controlled, authorized environments.
- [DATA_EXFILTRATION]: Includes templates for testing file reads using the
document()function (e.g.,/etc/passwd) and SSRF capabilities. These are documented testing procedures and do not perform unauthorized exfiltration from the agent's environment. - [COMMAND_EXECUTION]: Provides templates for testing Remote Code Execution (RCE) via Java extensions (
Runtime:exec), .NET script blocks (msxsl:script), and PHP functions. These are presented as payloads for testing target systems, not for execution against the agent's host. - [REMOTE_CODE_EXECUTION]: The skill describes methods for achieving RCE through XSLT processors. The provided payloads are standard security testing artifacts.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-influenced XSLT inputs for testing purposes. It correctly identifies this as an attack surface and provides guidance on boundary markers and processor fingerprints.
- [SAFE]: References to external tools and projects like Burp Suite, OWASP ZAP, and PayloadsAllTheThings are standard industry references and are used appropriately within the context of security auditing.
Audit Metadata