The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied data, specifically domain names and email addresses, that are interpolated directly into API request templates and shell commands.
Ingestion points: User inputs for variables such as {domain} and {email} are processed throughout SKILL.md (e.g., in sections §7.1, §8.1, and §11.2).
Boundary markers: The skill lacks explicit boundary markers or instructions to the agent to sanitize or validate these inputs before they are interpolated into executable contexts.
Capability inventory: The skill utilizes curl and PowerShell commands that are executed through the agent's shell access to perform network-based discovery.
Sanitization: No sanitization, escaping, or schema validation of the user-provided strings is implemented before they are used in generated command templates.
[DYNAMIC_EXECUTION]: The skill generates shell command templates (Bash and PowerShell) based on user-provided inputs, which are intended for execution in a command-line environment to perform reconnaissance tasks.

identity-provider-recon