liveavatar-demo
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFE
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill requests API keys for LiveAvatar, Anthropic, and LiveKit. However, it explicitly instructs the user to store these secrets in
.env.localfiles (which are gitignored in the reference repositories) and to use editor tools rather than shell commands to write them. This demonstrates a clear commitment to secret management best practices. - [EXTERNAL_DOWNLOADS]: The skill clones official repositories from the vendor (heygen-com) hosted on GitHub and installs packages using standard package managers (npm, uv/pip, brew). These actions are performed to provision the environment for the requested demonstrations.
- [INDIRECT_PROMPT_INJECTION]: The AI Sales Agent demo involves processing visitor input. The skill implementation includes basic sanitization by stripping YAML frontmatter and HTML comments from prompt components before they are sent to the underlying LLM.
- [COMMAND_EXECUTION]: The skill executes local commands for environment checks (e.g., node, git, uv versions) and process management (e.g., npx, uv run). These commands are restricted to setting up and running the specific demo applications and do not involve arbitrary or obfuscated shell execution.
Audit Metadata