The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing the codex exec CLI and other shell utilities such as ls, find, cp, and python. It provides command templates for image generation and post-processing tasks like background removal.
[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the @openai/codex package via NPM and the pillow library via PyPI. These are standard packages from established repositories associated with well-known technology organizations.
[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection via user-controlled data.
Ingestion points: User-provided image prompts are interpolated into instructions and shell commands within SKILL.md.
Boundary markers: There are no explicit delimiters or instructions provided to separate the user-generated content from the system instructions or to warn against embedded commands.
Capability inventory: The skill utilizes shell command execution (codex exec), file system access (/tmp/, ~/.codex/), and package installation tools (npm, pip).
Sanitization: User input is directly placed into a quoted string in a shell command (codex exec ... "<prompt instructions>"). If the input contains shell metacharacters like backticks or subshell expansions, it could potentially lead to arbitrary command execution within the user's environment.

codex-imagegen