The Agent Skills Directory

REMOTE_CODE_EXECUTION (CRITICAL): The skill documentation provides a command curl -fsSL https://cli.inference.sh | sh which downloads and executes a script from an untrusted external domain. This is a high-risk pattern that can lead to full system compromise.
EXTERNAL_DOWNLOADS (HIGH): The skill relies on a non-standard CLI tool and third-party models/apps hosted at inference.sh, which is not a verified or trusted source according to the security policy.
COMMAND_EXECUTION (MEDIUM): The skill explicitly allows the Bash(infsh *) tool, granting the agent the ability to execute any sub-command of the unverified infsh CLI, which interacts with remote infrastructure.
INDIRECT_PROMPT_INJECTION (LOW): The skill ingests untrusted data in the form of external image URLs and processes JSON output from tools.
Ingestion points: The --input flag accepts user-provided or externally sourced image URLs (e.g., https://your-image.jpg).
Boundary markers: Absent. There are no delimiters or warnings to ignore instructions embedded in the metadata of processed images or tool outputs.
Capability inventory: The skill can execute shell commands via infsh and write output to local files (e.g., > image.json).
Sanitization: None detected. External input is passed directly to the CLI tool.

image-upscaling