enrich-qa-test

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it parses user-provided QA test files to extract URLs and UI element names for use in automated browser actions. A maliciously crafted test file could attempt to redirect the agent to sensitive internal resources or include instructions within element names intended to manipulate the agent's behavior.
  • Ingestion points: The skill reads and parses the content of the provided (SKILL.md, Phase 1).
  • Boundary markers: The execution instructions do not specify any delimiters or safety markers to differentiate between instructions and data within the parsed file.
  • Capability inventory: The skill utilizes browser_navigate, browser_snapshot, browser_take_screenshot, and file system write operations (if --update is used) (SKILL.md, Phase 2 & 4).
  • Sanitization: The skill description does not include steps for URL validation or sanitization of element names before they are used as parameters for browser tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 06:50 AM