The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill's instructions and the generate_image.py script support passing the MODELSCOPE_API_KEY as a command-line argument using the --api-key flag. Passing sensitive credentials via CLI arguments is insecure because they can be exposed in process listings, shell history, and system logs.
[EXTERNAL_DOWNLOADS]: The skill requires users to clone and run an external repository from the author's GitHub account (github.com/gusibi/moli-tutu) to provide a local proxy service for R2 uploads. Additionally, the generate_image.py script performs network requests to download generated images from URLs returned by the ModelScope API.
[COMMAND_EXECUTION]: The skill workflow involves executing Python scripts that perform file system and network operations. Specifically, these scripts create directories for image storage, save binary image data, and programmatically update user Markdown files to include new image links.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes user-provided article content to extract themes and generate prompts for image creation, creating a surface for indirect prompt injection.
Ingestion points: Content from local Markdown files (path/to/article.md) or text pasted directly into the agent's context.
Boundary markers: The skill does not use specific delimiters or protective instructions to separate the article data from the agent's operational instructions during analysis.
Capability inventory: The skill has the ability to write local files (prompts and images), modify existing Markdown articles, and communicate with external APIs and local network services.
Sanitization: There is no evidence of content validation or sanitization applied to the article text before it is used to influence the image generation process.

article-image