grok

Warn

Audited by Snyk on Jul 3, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (high risk: 0.85). The required runtime path is that grok::run/grok::start spawns the host grok CLI, which streams Grok-generated event text back into grok::events/agent::events; this is outsider-authored free text because it originates from the xAI Grok service (not the operating user) and is ingested into the agent’s LLM context via the streamed event frames.

Issues (1)

W011
MEDIUM

Third-party content exposure detected (indirect prompt injection risk).

Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 3, 2026, 09:14 AM
Issues
1
Security Audit — snyk — grok