The Agent Skills Directory

[DATA_EXFILTRATION]: The script scripts/generate_image.py automatically attempts to load environment variables from a .env file in the current working directory and from ~/.env in the user's home directory. While this is a common practice for managing API keys, accessing these files exposes potentially sensitive credentials stored on the system to the agent context.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied image generation prompts and input images, which are passed to the OpenRouter API. The resulting output is written to the local filesystem. This creates a surface for indirect prompt injection where malicious instructions embedded in data could influence agent behavior or compromise output integrity.
Ingestion points: Untrusted data enters the agent context through the --prompt and --input-image arguments in scripts/generate_image.py.
Boundary markers: The skill does not implement delimiters or explicit instructions to ignore embedded commands within the prompt data.
Capability inventory: The skill possesses the capability to perform network operations (POST requests to OpenRouter) and write binary files to the local disk via Path.write_bytes.
Sanitization: There is no evidence of sanitization, validation, or escaping of the input prompt before it is interpolated into the API request body.

openrouter-img