e2e-test
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill defines a standard architecture for UI testing. All integrated tools and workflows (such as preflight checks, timeouts, and teardown logic) are appropriate for the stated goal of validating user journeys and generating failure reports.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface where untrusted web content is read via
mcp__claude-in-chrome__read_page(SKILL.md) to facilitate assertions. - Ingestion points:
mcp__claude-in-chrome__read_pageused for text and URL assertions. - Boundary markers: Not explicitly defined in logic snippets.
- Capability inventory: Full browser control (navigation, input, interaction) and state persistence via memory tools.
- Sanitization: Not present in example code snippets. While this ingestion creates a theoretical surface for indirect prompt injection, it is the fundamental mechanism required for E2E testing and is considered an expected functional behavior within this context.
Audit Metadata