The Agent Skills Directory

[Command Execution] (LOW): The script scripts/pdf_to_images.py executes the magick or convert binaries using subprocess.run. It correctly passes arguments as a list rather than a shell string, effectively mitigating shell-level command injection risks.
[Privilege Escalation] (MEDIUM): The SKILL.md documentation provides an instruction for users to execute a sudo sed command to modify the host's ImageMagick security policy (/etc/ImageMagick-*/policy.xml). This change relaxes default security constraints to allow PDF processing, which can expose the system to known historical vulnerabilities in ImageMagick's PDF parsing libraries.
[Indirect Prompt Injection] (LOW): The skill ingests untrusted PDF files which can serve as a vector for indirect prompt injection if the resulting images contain instructions that influence subsequent agent reasoning.
Ingestion points: The pdf_file argument in scripts/pdf_to_images.py used as input for conversion.
Boundary markers: Absent; no delimiters or "ignore instructions" warnings are applied to the processed data.
Capability inventory: System command execution and file system read/write access.
Sanitization: Basic validation is limited to checking the file existence and the .pdf extension.

pdf-to-images