how-to-make-ai-video
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEDATA_EXFILTRATIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [DATA_EXPOSURE]: The skill instructs the agent to read sensitive credential information from the local file system at
~/.varg/credentials. - [DATA_EXFILTRATION]: Extracted API keys are sent via HTTP Authorization headers to the vendor's API at
api.varg.ai. This represents functional behavior for the vendor's platform but involves handling of sensitive data. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill fetches Markdown and JSON content from
varg.aiandapi.varg.aito determine agent actions (SKILL.md). - Boundary markers: No delimiters or safety instructions are used to separate fetched content from the agent's core instructions.
- Capability inventory: The skill has access to
Bash(curl) andReadtools, which could be abused if remote content contains malicious instructions. - Sanitization: No explicit sanitization or validation of the remote content is described before the agent processes it.
- [DYNAMIC_EXECUTION]: The agent is instructed to install and run third-party skills (using
npx skills add) based on installation commands dynamically retrieved from thevarg.aisearch results at runtime. - [COMMAND_EXECUTION]: The skill frequently uses the
Bashtool to performcurloperations for searching, fetching guides, and publishing data.
Audit Metadata