The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run a local Python script, scripts/check_cover.py, to perform image validation and generate previews. This shell-based execution involves file paths constructed using user-provided data (e.g., the topic of the article), which creates a potential command injection vulnerability if the agent fails to sanitize or escape shell arguments properly.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted article content to extract headlines and keywords, which are then interpolated into prompts for the image_gen tool.
Ingestion points: Raw article content read in the first step of the workflow defined in SKILL.md.
Boundary markers: None identified. The instructions do not define delimiters or warnings to ignore instructions embedded in the ingested text.
Capability inventory: The skill uses the image_gen tool and executes local shell commands.
Sanitization: There is no evidence of filtering or sanitization of the article content before it is used in prompt generation.
[EXTERNAL_DOWNLOADS]: The script scripts/check_cover.py requires the Pillow library for image processing. While this is a well-known and reputable Python package, the skill documentation prompts for its installation via pip if it is not already present in the environment.

space-gzh-cover