vibecode-saas-app
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill creates a vulnerability surface where user-provided business context is directly used to generate application source code.
- Ingestion points: Untrusted data enters the context in
SKILL.mdunder "Step 2: CONTEXT", where the user provides problem descriptions, user personas, and feature requirements. - Boundary markers: Absent. The instructions in "Step 5: BUILD" command the agent to "CODE EXACTLY according to Blueprint" without advising the agent to ignore potentially malicious instructions embedded in the user's business context.
- Capability inventory: The agent performs extensive file-system write operations to create the project structure (Next.js components, layouts, and logic) as defined in the "CODER PACK" section.
- Sanitization: Absent. There are no explicit instructions for the AI to sanitize or validate the user's input before translating it into executable code.
- [COMMAND_EXECUTION]: The skill provides instructions for the user to execute shell commands to set up the generated environment.
- Evidence: The "AFTER COMPLETION" section in
SKILL.mdinstructs the user to runnpm installandnpm run devto start the application.
Audit Metadata