experiment-audit
Warn
Audited by Snyk on Jun 26, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow pulls experiment/flag/event data via the PostHog MCP (e.g.,
experiment-list, flag/metric/event queries in Steps 1–5), and those tool responses are customer-authored free-text fields (experiment names, variant names, event properties like$current_url) that the LLM ingests into context—an outsider source because it originates from the customer’s project data, not the operating user’s own authored prompts.
Issues (1)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
Audit Metadata