The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the mediaio CLI to perform core functions such as image/video generation (generate create), file uploads (upload create), and account status checks (account status). It also utilizes standard shell utilities like mktemp, grep, head, and file for temporary file management and result processing.
[EXTERNAL_DOWNLOADS]: The skill downloads generated assets from Media.io's storage services. It includes specific instructions to use mediaio generate download to ensure file integrity and avoid manual handling of signed URLs which contain storage credentials.
[DATA_EXFILTRATION]: By design, the skill reads local image and video files provided by the user and uploads them to the Media.io production servers for processing. This is the primary intended function of the skill.
[SAFE]: The skill demonstrates high security awareness by implementing multiple defensive measures:
Network Approval Gates: It mandates the use of host-native network approval mechanisms before any networked process is launched.
File Access Authorization: It explicitly requires the agent to request user permission before reading files located outside the active workspace.
Credential Safety: It manages authentication through the official CLI's browser-based login flow (mediaio auth login) and provides warnings against exposing access tokens or raw debug payloads.
URL Integrity: It includes a 'Result URL guardrail' to prevent the corruption of high-entropy signed storage URLs, emphasizing tool-based retrieval over manual copying.
User Confirmation: It requires explicit user approval for credit-spending operations and CLI upgrades, ensuring transparency in costs and system changes.

mediaio-generate