The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Insecure SSL Configuration. The download_url function in scripts/generate_image.py explicitly disables SSL certificate verification by setting ssl.CERT_NONE. This practice makes the connection vulnerable to man-in-the-middle (MITM) attacks, allowing a network-level attacker to intercept or replace the downloaded image data.
[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill ingests and processes prompts from external Markdown files provided via the --prompts-file parameter. If the input file is sourced from untrusted data (e.g., user-submitted content or web-scraped articles), it can be used to inject malicious instructions into the image generation pipeline.
Ingestion points: extract_prompts_from_markdown and load_image_plan_from_markdown in scripts/generate_image.py read content from files passed via the --prompts-file argument.
Boundary markers: The script uses headings (e.g., ##) and code blocks as delimiters, but lacks explicit "ignore instructions" safety boundaries for the extracted prompt text.
Capability inventory: The script has the capability to perform network requests (image generation APIs) and write to the local file system (saving images and modifying existing Markdown files via insert_images_to_markdown).
Sanitization: The script performs no sanitization or validation of the prompt text extracted from the Markdown files before passing it to the API providers.

image-gen