authoring-insight-alerts
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill features an AI detector that evaluates data retrieved from external database queries and insights, which represents a potential surface for indirect prompt injection if the source data contains malicious instructions.\n
- Ingestion points: Data retrieved via
posthog:insight-getand user-defined text in theinstructionsfield of thellmdetector configuration (SKILL.md).\n - Boundary markers: The skill does not describe or implement specific delimiters or boundary markers to separate data from instructions during the AI evaluation process.\n
- Capability inventory: The skill triggers automated notifications to Slack workspaces and email addresses via the
posthog:alert-destinations-createtool.\n - Sanitization: There is no documented sanitization or filtering of the insight data or user instructions before they are processed by the AI detector, beyond manual guidance to avoid sensitive data.
Audit Metadata