The Agent Skills Directory

[COMMAND_EXECUTION]: The helper script cv.py executes system commands tectonic and pdftoppm to compile LaTeX sources into PDFs and render them for visual inspection. The script implements security controls by validating input slugs with a strict regular expression (^[A-Za-z0-9._-]+$), which prevents path traversal and command injection.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, specifically job descriptions (JDs), to tailor resumes. The skill provides clear operational constraints, such as the requirement to audit all claims against a durable profile.md and explicit instructions to never invent content, which serve as mitigations against instructions embedded in external documents.
[DATA_EXPOSURE]: This skill manages sensitive personal information including contact details, work history, and location stored in profile.md. The skill follows a repository contract that segregates durable career data from disposable tailored packets (stored in a gitignored tmp/ directory), reducing the risk of accidental exposure or leakage into version control.
[DATA_EXFILTRATION]: The skill's intended purpose includes submitting career documents to external web forms (e.g., job boards) via browser automation. This involves sending sensitive data to third-party domains; however, the skill provides guidance on using browser automation tools for this purpose, including manual checkpoints and screenshot verification for the user.

cv-creator