The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text for speech synthesis and media from external URLs, creating a surface for potential injection.
Ingestion points: The script input in SKILL.md and external media URLs (social links, public URLs) described in references/media-input.md.
Boundary markers: None present for the text-based script input.
Capability inventory: The skill utilizes create-lipsync, upload-media, and shell-based curl commands for network and file operations.
Sanitization: No explicit sanitization or filtering is described for the script or external URL content.
[COMMAND_EXECUTION]: The documentation provides explicit shell commands for users or agents to perform file operations and network requests.
Evidence: references/api.md includes curl -sS --fail -X PUT -H 'Content-Type: video/mp4' --data-binary @./clip.mp4 "<uploadUrl>" and SKILL.md suggests curl -L -o clip_lipsync.mp4 "<url>" for downloading results.
[EXTERNAL_DOWNLOADS]: The skill is designed to fetch generated media content from external infrastructure.
Evidence: The workflow culminates in fetching an outputMediaUrl provided by the Sync.so API, which is a vendor resource.

lipsync