filler-removal
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation instructs the user to clone the
agamm/video-agentrepository from GitHub to access the necessary scripts and CLI tools. This is a functional requirement for the skill's operation within its intended environment. - [COMMAND_EXECUTION]: The instructions guide the agent to execute shell commands using
uv runfor thevideo-agentutility andffmpegfor video manipulation. These commands are standard for video processing and are used to transcribe audio, analyze speech segments, and perform video concatenation. - [INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted data by transcribing video audio into text. While this text is then processed by the agent to identify filler words, the risk of indirect prompt injection is minimal due to the skill's focus on specific keyword identifiers (e.g., 'um', 'uh') and the inclusion of manual verification steps, such as reviewing 3-second splice previews.
Audit Metadata