The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill is designed to process untrusted external content (resume data in JSON format) and produce formatted PDF output. This creates a surface where malicious instructions could be embedded in the input data to influence the agent or downstream consumers of the PDF.
Ingestion points: ResumeBuilder.from_json() method and the --input CLI argument.
Boundary markers: The documentation does not specify the use of delimiters or instructions to ignore embedded content.
Capability inventory: The skill has the capability to write to the file system (.save() method).
Sanitization: There is no mention of input sanitization or validation schema in the documentation.
Missing Implementation (LOW): The core logic file scripts/resume_builder.py is referenced throughout the SKILL.md file but is absent from the provided skill folder. This prevents a complete audit of the actual code behavior, including potential hidden network calls or unsafe parsing logic.
Dependency Review (SAFE): The skill relies on reportlab and Pillow. The versions specified (>=4.0.0 and >=10.0.0) are modern and do not have known critical vulnerabilities at the time of analysis.

resume-builder