The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands to manage files and compile resumes.
It uses which xelatex to verify environment prerequisites.
It invokes make en and make zh to compile LaTeX source into PDF documents.
The provided Makefile uses cat and sed to dynamically generate localized versions of the resume source file.
[DATA_EXFILTRATION]: While the skill primarily processes data locally within the user's workspace, it handles sensitive personal information (name, email, phone, work history).
The data is stored in plain text markdown files within the experiences/ directory.
There is no evidence of unauthorized external transmission, but users should be aware that their personal data is written to the local file system by the agent.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect injection surface by processing untrusted data (user-provided resumes or job descriptions) and inserting it into an executable context (LaTeX source).
Ingestion points: User-provided text via $ARGUMENTS in the init, generate, and add subcommands, and content read from external job description files.
Boundary markers: None explicitly defined in the instructions for separating user data from LaTeX templates.
Capability inventory: File writing (resume.tex), shell execution (xelatex, make, sed).
Sanitization: The instructions mention basic escaping for & and %, but do not specify escaping for the LaTeX command character \ or grouping characters { }, which could allow malicious input to execute LaTeX commands during the build process.

resume