create-subagent
Pass
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands such as
git rev-parse,mkdir, andln -s. These operations are used to manage the local repository structure and are consistent with the skill's stated purpose of scaffolding files. - [PROMPT_INJECTION]: The skill constructs an
AGENTS.mdfile by interpolating user-provided text into a template. This represents an indirect injection surface if the generated content is later used to define agent behavior without further review. - Ingestion points: User input captured by the script
scripts/gather-subagent-info.cjsvia terminal prompts. - Boundary markers: None; the gathered text is placed directly into markdown headers and workflow sections.
- Capability inventory: Local file system management, directory scaffolding, and symlink creation.
- Sanitization: Minimal; basic string trimming is performed on input values.
Audit Metadata