exploring-llm-costs
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted event data from the PostHog platform to generate insights and alerts.
- Ingestion points: Event properties and custom dimensions are retrieved via the
posthog:execute-sqlandposthog:query-llm-tracetools as seen in SKILL.md and breakdown-patterns.md. - Boundary markers: The skill does not implement explicit boundary markers or instructions to ensure the agent ignores potential natural language instructions embedded within event property values.
- Capability inventory: The agent is empowered to execute HogQL queries, create saved insights using
posthog:insight-create, and configure automated alerts viaposthog:alert-create. - Sanitization: There are no explicit steps for sanitizing or validating external property values before they are incorporated into the agent's context or persisted in insights.
Audit Metadata