app-validation

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [Browser Security Configuration]: The skill instructs the agent to disable specific browser security features, such as LocalNetworkAccessChecks, via a configuration file (.playwright-config.json). This is a documented requirement to allow the production Fabric portal to communicate with an application hosted on localhost during development. While this reduces browser isolation, it is scoped to the specific testing session.
  • [Persistent Session Handling]: The instructions recommend using --persistent browser profiles to manage Azure Active Directory (AAD) authentication. This allows the user to sign in manually once while the agent maintains the session across subsequent runs. This is a standard approach for testing authenticated workflows, though users should be aware that the profile directory will store session-related data like cookies.
  • [Dynamic Browser Execution]: The skill utilizes page.evaluate and run-code to execute JavaScript logic directly within the browser context. This is used for diagnostic purposes, such as walking the frame tree and performing visual consistency checks on charts and layouts.
  • [Indirect Prompt Injection Surface]: The skill is designed to ingest and process data from the browser, including console messages and the accessibility tree (snapshot). This represents a potential surface where external content could influence agent behavior. However, the skill includes specific filters to prioritize messages originating from the local application (http://localhost:5173) while ignoring noise from the external portal.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 12:19 PM
Security Audit — agent-trust-hub — app-validation