The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads Node.js binaries and SHA256 checksum files from the official nodejs.org distribution server. This is a well-known service and the skill includes a validation step to verify file integrity using the provided checksums.
[REMOTE_CODE_EXECUTION]: The skill provides instructions to install and execute the '@iqinghu/qhkit' package from the npm registry. It also facilitates the installation of the Node.js runtime environment to support this tool.
[COMMAND_EXECUTION]: The skill executes various shell commands including 'npm install', 'tar', and the 'qhkit' utility itself to perform its primary function of image cloning.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied image paths and descriptive prompts which are then incorporated into command-line arguments, creating a potential surface for indirect prompt injection.
Ingestion points: User-provided local image filenames and descriptive prompts are used as inputs for the 'qhkit image generate' command in SKILL.md.
Boundary markers: The command templates do not utilize explicit delimiters or warnings to isolate or ignore potentially malicious instructions embedded in the user-provided prompts.
Capability inventory: The skill relies on the 'qhkit' CLI, which performs network requests to external image generation APIs and accesses the local file system for source images.
Sanitization: There is no evidence of sanitization or validation of the user-supplied strings before they are embedded into the JSON payload used by the command-line tool.

qinghu-main-image-clone