The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to verify the existence of configuration files in the user's home directory or project directory using test -f. It also uses npx to trigger image generation scripts.
[DYNAMIC_EXECUTION]: The workflow involves executing a local TypeScript script (scripts/main.ts) using the bun runtime via npx. This script is used to handle the assembly of image prompts and the generation process as described in the prompt-assembly.md guide.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user content, such as articles or social media posts, to generate infographic outlines and subsequent image prompts. This creates a surface where malicious instructions embedded in the input text could attempt to influence the image generation output or the agent's reasoning during the analysis phase.
Ingestion points: Untrusted source content provided by the user, saved to files such as source.md in the generated session directory.
Boundary markers: The skill utilizes a structured prompt assembly template defined in references/workflows/prompt-assembly.md to wrap the content, though it lacks explicit "ignore embedded instructions" warnings for the image generation model.
Capability inventory: File system write operations, shell command execution via npx, and invocation of external image generation tools.
Sanitization: No specific sanitization or filtering logic is documented for the user-provided content before it is interpolated into the final generation prompts.

aiset-xhs-images