The Agent Skills Directory

[DATA_EXFILTRATION]: The skill includes an example command designed to access sensitive cloud metadata services (http://169.254.169.254/latest/meta-data/) to test for Server-Side Request Forgery (SSRF) vulnerabilities. While expected in a security testing context, this pattern involves accessing internal data structures.
[EXTERNAL_DOWNLOADS]: The documentation provides an example for downloading a binary file (tool.exe) from a remote IP address (10.10.10.1). While typical for security tooling, downloading executables from external sources carries inherent risks if the source is not authenticated.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a wrapper for the curl tool, accepting arbitrary URLs and headers as input. This creates an attack surface where a malicious endpoint could return instructions intended to influence the agent's next steps.
Ingestion points: URLs and API paths provided as inputs in SKILL.md.
Boundary markers: None identified; the skill does not explicitly instruct the agent to ignore instructions embedded in the retrieved content.
Capability inventory: The skill uses curl to perform network operations and can download files to the local system.
Sanitization: No explicit sanitization or validation of the retrieved content is described.

curl