The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to construct and execute shell commands (specifically 'curl', 'base64', and 'mkdir') using user-provided inputs such as the image prompt and reference URLs. The lack of instructions for sanitizing or escaping these inputs makes the skill vulnerable to command injection if a user provides input containing shell metacharacters like single quotes, backticks, or dollar signs.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to 'api.labnana.com' to submit generation tasks and retrieve image data. This involves transmitting the 'LISTENHUB_API_KEY' environment variable to an external service that is not part of the trusted vendor list.
[PROMPT_INJECTION]: The skill exhibits risk factors for indirect prompt injection. Ingestion points: User-provided free-text prompts (Step 1) and reference image URLs (Step 4). Boundary markers: None are specified to isolate untrusted user content within the shell command context. Capability inventory: Access to shell tools ('curl', 'mkdir', 'echo', 'base64') and file system writes. Sanitization: There are no instructions to escape or validate external content before it is interpolated into the workflow.

image-gen