managing-subscriptions

Pass

Audited by Gen Agent Trust Hub on Nov 5, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill manages 'AI prompt reports' which store and process up to 4,000 characters of arbitrary text used by the AI planner to generate reports.
  • Ingestion points: The prompt field in the ai_prompt resource type, accessed via posthog:subscriptions-retrieve and posthog:subscriptions-list (in references/subscription-types.md).
  • Boundary markers: The instructions do not define specific delimiters to separate the stored prompt content from the agent's internal instructions during retrieval or update operations.
  • Capability inventory: Tools like posthog:subscriptions-create and posthog:subscriptions-partial-update allow writing arbitrary text into the PostHog backend, which is later processed by an AI planner.
  • Sanitization: No sanitization or filtering logic is specified for the prompt text beyond a character limit.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 5, 2026, 10:27 AM