huashu-motion
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEDYNAMIC_CONTEXT_INJECTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [DYNAMIC_CONTEXT_INJECTION]: The skill contains logic in
SKILL.mdthat instructs the agent to perform an automated version check by executinggit ls-remoteagainst a remote repository to compare local and remote tags. - [INDIRECT_PROMPT_INJECTION]: The skill's primary workflow involves ingesting untrusted data from external URLs (via
om grab) and user comments (viaom inbox) to modify the project's source code (index.html). This creates a significant surface for indirect prompt injection attacks where malicious data could influence the agent's code generation. - [EXTERNAL_DOWNLOADS]: The skill's helper script
bin/omis configured to download and execute thehuashu-motionpackage from the NPM registry usingnpx. It also provides commands to install external asset packs, such as real-instrument samples viaom samples build. - [COMMAND_EXECUTION]: The agent is instructed to manage local development environments, including starting a local server on
127.0.0.1for the 'studio' preview and invoking rendering tools likeffmpegfor video export.
Audit Metadata