The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided content which presents a surface for indirect prompt injection.
Ingestion points: External data enters the system through the --prompt and --input parameters within the tools/ltx2.py script as documented in SKILL.md.
Boundary markers: The skill documentation does not outline the use of specific boundary markers or delimiters to differentiate between instructions and data.
Capability inventory: The skill's capabilities are limited to media generation (.mp4 files), local file processing via ffmpeg, and interaction with the user's Modal account. No sensitive file access or arbitrary network exfiltration capabilities were found.
Sanitization: No specific input sanitization or validation measures are described for the text or image inputs.
[COMMAND_EXECUTION]: The skill provides instructions for executing various CLI tools and scripts to perform its tasks.
The documentation lists commands for running local Python scripts (ltx2.py, upscale.py, flux2.py, qwen3_tts.py) and standard utilities like ffmpeg.
It includes commands for deploying applications and managing secrets on the Modal platform (modal deploy, modal secret create). These operations are transparent, well-documented, and consistent with the functional purpose of the skill.

ltx2