The Agent Skills Directory

[COMMAND_EXECUTION]: The shell scripts utilize system utilities such as curl for network requests, jq for parsing responses, and python3 for safe JSON encoding. It also leverages a muapi CLI tool which is expected for interacting with the service.
[EXTERNAL_DOWNLOADS]: The skill communicates exclusively with api.muapi.ai to facilitate image and video processing tasks. These interactions are core to the skill's functionality and trace back to the service's official infrastructure.
[DATA_EXFILTRATION]: The skill allows users to upload local files to the MuAPI cloud for processing. This behavior is the intended and transparent method for the service to operate on user-provided assets.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied prompts and media URLs, which serves as a potential surface for indirect injection.
Ingestion points: Input is accepted via command-line arguments across the edit-image.sh, enhance-image.sh, lipsync.sh, and video-effects.sh scripts.
Boundary markers: Prompts are structured within JSON payloads during API transmission, providing a clear delimiter between the command structure and user data.
Capability inventory: The skill possesses file-reading capabilities for uploads and network access to the vendor's API.
Sanitization: All scripts use a Python-based JSON encoder to ensure that user prompts are correctly escaped before being sent to the remote API, effectively mitigating basic structural injection attempts at the communication layer.

muapi-media-editing