project-scaffolding
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill follows security best practices, such as providing
.env.exampletemplates and warning against committing secrets to version control.\n- [EXTERNAL_DOWNLOADS]: Thescripts/refresh_versions.pyscript fetches version information from official registries (registry.npmjs.organdpypi.org). These are well-known, trusted services used to maintain theassets/versions.jsonmanifest.\n- [INDIRECT_PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection by interpolating user-supplied strings into generated project files.\n - Ingestion points: The
ProjectConfigobject inscripts/scaffold.pyreceives user input forname,description, andauthor.\n - Boundary markers: None are present in the templates to separate user strings from the surrounding code or markdown.\n
- Capability inventory:
scripts/scaffold.pyhas the capability to write numerous file types (JSON, YAML, Python, TypeScript, etc.) and create directory structures.\n - Sanitization: No sanitization or escaping of the user-provided strings was observed before interpolation into templates.\n- [REMOTE_CODE_EXECUTION]: The skill recommends standard framework initialization commands (e.g.,
npx create-next-app,npm create vite). These commands fetch code from official package registries, which is standard industry practice for project bootstrapping.
Audit Metadata