full-stack-development

Pass

Audited by Gen Agent Trust Hub on Oct 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data to drive its development lifecycle, creating a surface for potential indirect prompt injection attacks.
  • Ingestion points: The skill ingests raw user-supplied goals, corrections, and requirement changes, as well as external design references such as TailwindUI material and Pencil/Figma exports.
  • Boundary markers: There are no instructions for the agent to use specific delimiters or protective markers when these untrusted inputs are incorporated into the prompt context.
  • Capability inventory: The skill possesses significant capabilities, including writing workspace files, generating backend and frontend code, and executing shell commands for testing, building, and visual auditing.
  • Sanitization: No mechanisms for automated sanitization, validation, or escaping of the user-provided inputs or external assets were found; the workflow relies primarily on user confirmation stages.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 13, 2026, 12:18 PM