mograph
Pass
Audited by Gen Agent Trust Hub on Nov 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The scripts
scripts/cut_plan.pyandscripts/make_spec.pyutilize thesubprocess.runfunction to invoke system tools likeffmpegandffprobefor media analysis and video editing. This is a primary function of the skill and arguments are passed as lists to prevent shell injection. - [EXTERNAL_DOWNLOADS]: The
templates/cover_from_base.example.pyscript automatically downloads image assets such as background templates and hair reference frames from the author's official AtomGit and GitHub repositories if they are missing locally. These downloads target specific binary assets from vendor-owned infrastructure. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided transcripts and口播稿 as data to determine where to place motion graphics and where to cut video segments. While this constitutes a data ingestion surface, the risk is mitigated by the use of deterministic regex matching and local processing.
- [DYNAMIC_EXECUTION]: The skill's architecture includes a "muscle layer" where the agent is instructed by
references/research-protocol.mdto search for and implement community motion graphics code (e.g., Remotion or HyperFrames components) at runtime. The protocol includes specific safety checks and scoring for licenses and code quality.
Audit Metadata