The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection vulnerabilities.
In g e s t i o n p o i n t s: The skill reads all source files, configuration files, and dependency manifests from a target directory (./audit-target) which may be populated from a user-provided GitHub repository.
Bo u n d a r y m a r k e r s: The instructions lack delimiters or explicit directives for the agent to disregard instructions found within the analyzed codebase.
Ca p a b i l i t y i n v e n t o r y: The skill uses gh for cloning, g r e p and g l o b for file discovery, and has file-read and file-write access within the workspace to generate reports.
Sa n i t i z a t i o n: No sanitization is performed on the content read from the target project before the agent processes it.
[EXTERNAL_DOWNLOADS]: The skill performs downloads from external sources.
E v i d e n c e: gh repo clone <url> ./audit-target in S K I L L . m d.
C o n t e x t: This operation is the intended primary purpose of the skill, allowing it to audit remote repositories as requested by a user.
[COMMAND_EXECUTION]: The skill utilizes several system commands for its operation.
E v i d e n c e: The skill invokes gh, g r e p, and m k d i r via the shell.
C o n t e x t: These commands are used for project reconnaissance, pattern matching during the security audit, and directory management for the final reports.

security-scanner