The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The load_env function in scripts/background_remove.py is designed to search for and read .env files from several sensitive locations, including the user's home directory (~/.env) and a global configuration path (~/.config/skills/.env). While intended for local configuration, these files often contain sensitive secrets unrelated to the skill's function.
[EXTERNAL_DOWNLOADS]: The skill utilizes the rembg library, which automatically initiates a download of the U2-Net model (approximately 170MB) from a remote repository upon its first execution.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by processing external image files whose paths are provided by the user.
Ingestion points: User-supplied image paths processed by the remove_background function in scripts/background_remove.py.
Boundary markers: No boundary markers or "ignore embedded instructions" warnings are implemented.
Capability inventory: The skill possesses file system write capabilities (saving processed PNG/WebP files) and subprocess execution for script invocation.
Sanitization: No sanitization or metadata stripping is performed on the input images before processing.

background-remove