The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes the codex exec command line tool to perform image generation. It also manages file system operations, such as moving or copying generated images from the ~/.codex/generated_images/ cache to user-defined output directories. Additionally, it references downstream execution of Python scripts for article publishing.
[EXTERNAL_DOWNLOADS]: The skill requires the codex CLI as a mandatory external dependency and provides installation instructions using npx to fetch the skill from the vendor's repository.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting untrusted article titles and content. These inputs are analyzed to derive visual parameters and are interpolated into prompts for the image generation backend.
Ingestion points: User-provided article content and titles processed within the generation workflow in SKILL.md.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are described for the prompt construction phase.
Capability inventory: File system write access for prompt and image files, and command execution via the codex exec utility.
Sanitization: There is no explicit mention of sanitizing or escaping user-provided text before it is used to generate the final image prompt.
[DATA_EXFILTRATION]: The skill reads configuration files from ~/.config/soia-skills/ and accesses ~/.codex/models_cache.json to identify supported model versions. While these operations are localized to the skill's functional requirements, they involve accessing tool-specific metadata and configurations.

soia-pkm-cover-image