threejs-game-artifact-cleanup
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands (PowerShell 7) to remove files using
Remove-Item -LiteralPath. It also invokes "re-generation commands" and build/test processes as part of the verification and cleanup workflow. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from untrusted sources, including
Work Items,HTML/CSS/JSONreferences, andapproval ledgers. This creates a vulnerability where malicious project files could influence the agent to delete unintended files or misinterpret file usage. - Ingestion points:
Work Itemfiles,HTML/CSS/JSONinternal references, and project configuration files (SKILL.md Step 2). - Boundary markers: The skill lacks explicit instructions or delimiters to ignore instructions embedded within the data it parses for references.
- Capability inventory: File deletion (
Remove-Item), PowerShell command execution, and build script invocation. - Sanitization: Employs absolute path resolution,
-LiteralPathfor PowerShell operations to prevent wildcard expansion, and root-directory boundary checks to prevent path traversal. - [DYNAMIC_EXECUTION]: The skill identifies and potentially executes "re-generation commands" and "build/test configurations" derived from the project environment (SKILL.md Step 3 and 6). Executing commands defined within project metadata or external configuration files constitutes a dynamic execution surface.
Audit Metadata