The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous shell command examples for security auditing. These include network probes via curl and automated scanning using specialized tools like dotdotpwn and wfuzz to detect path traversal vulnerabilities on remote targets.
[INDIRECT_PROMPT_INJECTION]: The skill outlines a methodology for processing external data (URL parameters, cookies, and server responses) that could contain malicious content designed to influence the agent's behavior during a security audit.
Ingestion points: The skill processes URL parameters (file=, path=, etc.), HTTP cookies, and dynamic server error messages (SKILL.md).
Boundary markers: No explicit boundary markers or "ignore embedded instructions" warnings are provided for the agent when interpreting external tool outputs.
Capability inventory: The skill facilitates network requests, automated fuzzing, and directory traversal probes through external CLI tools.
Sanitization: While the skill provides remediation patterns for the target application (e.g., path canonicalization), it does not specify sanitization for the agent's internal handling of the data it retrieves.
[EXTERNAL_DOWNLOADS]: The skill references the OWASP Web Security Testing Guide (WSTG) as a source for its methodology and framework, which is an established and trusted resource in the security community.

path-traversal