gsp-scaffold
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes system commands for environment setup and build verification.
- Evidence: Uses Bash to run
rm -rf,npx next build, andnpm run buildas part of the scaffolding process. - [EXTERNAL_DOWNLOADS]: Retrieves components and packages from official registries and established developer tools.
- Evidence: Fetches resources via
npm installand initialization tools for Next.js, shadcn/ui, and React Native Reusables. - [PROMPT_INJECTION]: Identifies a surface for indirect prompt injection where external data influences tool execution.
- Ingestion points: Reads project-specific manifests from
install-manifest.mdand configuration fromconfig.json(SKILL.md). - Boundary markers: Absent; content from manifest files is directly used to construct and execute shell commands.
- Capability inventory: Full Bash tool access, package manager execution, and file system modification capabilities.
- Sanitization: Absent; the skill executes component additions and dependency installs without verifying the legitimacy or safety of the entries provided in the manifest files.
Audit Metadata