app-builder

Pass

Audited by Gen Agent Trust Hub on Aug 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • Local Script Execution: The skill executes several Node.js scripts (e.g., build-model-app.js, preview-app.js, spec-lint.js) located within the plugin's root directory. These scripts perform core tasks such as linting application specifications, generating wireframes, and managing the build process.
  • Power Platform CLI Integration: The tool utilizes the pac command-line interface to interact with Dataverse environments. This includes tasks such as environment discovery, metadata retrieval, and the uploading of generative pages. These operations are standard for Power Apps development workflows.
  • Automated UI Code Generation: The skill dispatches a specialized headless worker agent to generate TypeScript React (.tsx) files for application pages. This automated generation is based on a plan derived from user input, which is the intended behavior of the skill.
  • Command Execution via Node: The skill includes an inline Node.js execution pattern (node -e) to perform schema validation (linting) on the App Spec file. This is used to ensure the integrity of the application design before deployment.
  • Environment Interaction: The skill reads and updates environment-specific resources, such as Dataverse tables, security roles, and sitemaps. These interactions are scoped to the authenticated user's Power Platform environment.
  • Indirect Prompt Injection Surface: As a tool that generates code and application logic from natural language, the skill inherently possesses a surface for indirect prompt injection. The author has implemented validation and promotion scripts (e.g., promote-intent-pages.js) to check the structural integrity of the generated output before it is committed to the application specification.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 9, 2026, 01:15 AM
Security Audit — agent-trust-hub — app-builder