yt-shorts
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a Python script located at
../yt-script/hookscore.py. While the path is relative, this constitutes the execution of code residing outside the immediate skill directory. - [DYNAMIC_EXECUTION]: The instruction to run a replacement line through the
hookscore.pyscript implies runtime execution of local code based on generated content. - [INDIRECT_PROMPT_INJECTION]: The skill processes video transcripts, which are external, untrusted data sources.
- Ingestion points: The
SKILL.mdfile instructs the agent to "Read the transcript" to identify potential clips. - Boundary markers: There are no instructions to use delimiters or specific ignore markers for the transcript content.
- Capability inventory: The skill has the capability to execute local scripts and generate formatted output for the user.
- Sanitization: No sanitization or validation of the transcript content is specified, making it possible for embedded instructions in the transcript to influence agent behavior.
Audit Metadata