vignette-catalog-scaffold
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
subprocessmodule inscripts/scaffold.pyto execute localgitcommands. These commands are used to initialize new repositories (git init) and resolve project roots (git rev-parse). The execution uses argument lists rather than shell strings, which is a safe practice. - [INDIRECT_PROMPT_INJECTION]: The skill includes an '--adopt' mode and instructions to reconcile existing repository files (such as
README.mdandpyproject.toml). This interaction with pre-existing, potentially untrusted local data creates a surface for indirect prompt injection. However, the skill primarily preserves these files rather than processing their content for execution, and it uses regex validation for catalog identifiers.
Audit Metadata