The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/validate_format.py script executes system utilities pdfinfo and pdffonts using subprocess.run. These calls are used to extract document metadata (page count, margins) and font information for compliance checking. The implementation correctly uses a list of arguments to avoid shell interpolation, which is a secure practice.
[INDIRECT_PROMPT_INJECTION]: The skill presents an indirect prompt injection surface through the scripts/customize_template.py script. This script takes untrusted user input (e.g., paper titles, author names, affiliations) and interpolates it directly into LaTeX templates. These files could potentially contain instructions that influence the agent when it subsequently reads or compiles the documents. Evidence includes:
Ingestion points: CLI arguments --title, --authors, --affiliations, and --email in scripts/customize_template.py.
Boundary markers: None; the script uses simple regex replacement of placeholders.
Capability inventory: The agent possesses Bash and Write capabilities, allowing it to generate and compile these LaTeX documents.
Sanitization: No validation, escaping, or filtering is applied to the user-supplied strings before they are written to the templates.
[SAFE]: No malicious code, exfiltration logic, or obfuscation techniques were detected. The skill's functionality aligns with its stated purpose, and all external references point to reputable scientific and academic domains.

alterlab-venue-templates