The Agent Skills Directory

[PROMPT_INJECTION]: The skill's instructions and reference files do not contain any attempts to override agent behavior, bypass safety guidelines, or extract system prompts. It follows a clear, task-oriented workflow.
[DATA_EXFILTRATION]: No evidence of unauthorized data access or network transmission to untrusted domains. Network activities are restricted to referencing a trusted homepage on GitHub, and file operations are confined to the local project directory for saving infographics and configuration.
[CREDENTIALS_UNSAFE]: There are no hardcoded API keys, tokens, or credentials. The skill manages user preferences through a local configuration file (EXTEND.md) without storing sensitive secrets.
[REMOTE_CODE_EXECUTION]: The skill does not include any scripts, binary executables, or package installation commands. It relies entirely on natural language instructions to invoke pre-existing image generation tools in the agent's environment.
[COMMAND_EXECUTION]: The skill performs benign file system operations, such as creating directories for the project and writing markdown files for prompts and analysis. These are standard and necessary for its documented functionality.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface as it processes untrusted user content to generate image prompts. While the prompt assembly uses markdown headers to separate sections, it lacks explicit delimiters or instructions to ignore embedded commands within the user's source text. However, the mandatory 'Smart Confirm' workflow (Step 2) serves as a robust human-in-the-loop mitigation, ensuring the user reviews the generated plan before images are produced.

baoyu-image-cards