product-inventor

Pass

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted user data to drive its core logic and code generation outputs.
  • Ingestion points: Untrusted data enters the agent context through command parameters in SKILL.md, specifically via /invent [ideia/produto], /diagnose [produto/descricao], and /polish [tela/produto].
  • Boundary markers: The instructions lack explicit delimiters or "ignore instructions" warnings to encapsulate and isolate user-provided product descriptions from the agent's internal logic.
  • Capability inventory: Across its workflows, the skill generates complex project structures (folders and files), UI components, and business logic. It utilizes tools such as claude-code, gemini-cli, and codex-cli which have the capability to write to the file system and interact with external APIs based on the agent's derived plan.
  • Sanitization: There is no evidence of sanitization, validation, or escaping of the user-provided descriptions before they are interpolated into the prompt or used to generate executable code.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 9, 2026, 07:20 AM