The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The installation instructions for Linux, macOS, and Windows recommend piping a remote script directly to the system shell using commands like curl https://genmedia.sh/install -fsS | bash and irm https://genmedia.sh/install.ps1 | iex. This is a high-risk pattern that executes unverified code from a non-whitelisted domain.
[REMOTE_CODE_EXECUTION]: The CLI features a genmedia update command that fetches and executes software updates from a remote server, as well as genmedia skills install which downloads and installs additional agent skills from an external registry into the agent's environment.
[COMMAND_EXECUTION]: The skill's primary purpose is to execute various CLI commands on the local system, providing a broad attack surface for command injection if input parameters are not correctly sanitized by the agent.
[EXTERNAL_DOWNLOADS]: The skill facilitates multiple external downloads, including the initial installer, CLI updates, and model-specific extensions from the genmedia.sh domain.
[DATA_EXFILTRATION]: The genmedia upload command allows for the transmission of local files to the fal.ai CDN. This functionality presents a risk of data exfiltration if the agent is instructed to upload sensitive system configuration or private user data.
[CREDENTIALS_UNSAFE]: The genmedia setup command handles sensitive API credentials (FAL_KEY) and offers to persist them in a local configuration file, making them potentially accessible to other processes on the system.

genmedia