exploring-autocapture-events

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides legitimate documentation and workflow guidance for using PostHog's autocapture feature. No evidence of obfuscation, remote code execution, or data exfiltration was detected.
  • [PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection (Category 8) by processing untrusted data from user-controlled web elements.
  • Ingestion points: Untrusted data enters the agent context via the events table, specifically through the elements_chain and elements_chain_texts columns (documented in SKILL.md and example-queries.md).
  • Boundary markers: Absent. The instructions do not specify any delimiters or warnings to the agent to ignore potentially malicious instructions embedded within the captured event data.
  • Capability inventory: The skill utilizes data reading through HogQL queries and action management via the posthog:action-create tool.
  • Sanitization: Absent. The skill handles captured data as raw strings for the purpose of regex matching and analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 04:13 AM