replay-ux-research

Pass

Audited by Gen Agent Trust Hub on Apr 24, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted data originating from external users, such as URL query parameters, search terms, and navigation breadcrumbs. This creates a surface for indirect prompt injection where an attacker could navigate to specific URLs or perform searches containing malicious instructions designed to manipulate the AI agent's analysis or output.
  • Ingestion points: In SKILL.md, the tools search_events, get_replay_details, and get_sentry_resource fetch user-controlled strings like URLs, search queries, and activity breadcrumbs.
  • Boundary markers: No specific delimiters or boundary markers are instructed to be used when interpolating this untrusted data into the agent's context.
  • Capability inventory: The skill uses search_events, get_replay_details, search_issues, and get_sentry_resource (all tools in SKILL.md). These are restricted metadata retrieval tools within the Sentry platform and do not include arbitrary command execution or file system writes.
  • Sanitization: The instructions include a requirement to anonymize user email addresses, but there is no mention of sanitizing or escaping the text content of URLs or breadcrumbs to prevent injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 24, 2026, 11:48 PM