project-init
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Node.js scripts (e.g.,
.claude/hooks/lib/session-init-helpers.cjs) to determine the state of the project and validate the project configuration. These operations are restricted to the local environment and are consistent with the tool's intended purpose of project initialization. - [INDIRECT_PROMPT_INJECTION]: The skill processes project files such as
CLAUDE.mdandAGENTS.md. Although these files could theoretically contain malicious instructions (indirect prompt injection), the skill implements 'smart-merge' protocols to prevent blind overwriting of user-authored content and maintains a clear distinction between project-neutral and project-specific instructions. - [DYNAMIC_EXECUTION]: The skill utilizes
node -eto run short evaluation scripts for state checking. This is used for internal project auditing and verification rather than executing untrusted external input.
Audit Metadata