The Agent Skills Directory

[SAFE]: The skill is designed for local processing of repository content to create professional documents. It does not exhibit malicious behavior such as data exfiltration or unauthorized network access.
[COMMAND_EXECUTION]: The instructions direct the agent to check for the presence of pandoc using which and to verify PDF extraction with pdftotext. These are standard, safe operations for the skill's purpose.
[EXTERNAL_DOWNLOADS]: The reference material includes documentation on how a user can manually install necessary fonts via the system package manager (dnf), which is a legitimate setup activity.
[PROMPT_INJECTION]: The skill processes repository files as input. Although this is a theoretical vector for indirect prompt injection, the skill's structured extraction process and rigorous quality gates act as significant mitigations.

cv-builder