exploring-llm-evaluations
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill documents and interacts with official PostHog AI observability tools. All referenced tools (e.g.,
posthog:llma-evaluation-list,posthog:execute-sql) are internal to the PostHog platform. - [SAFE]: No external network requests, suspicious downloads, or obfuscated code were detected in the instructions.
- [SAFE]: The SQL queries provided are for analytical purposes (HogQL) over
$ai_evaluationand$ai_generationevents and do not target sensitive system tables or credentials. - [SAFE]: The skill uses a well-defined YAML frontmatter and documentation style consistent with a legitimate developer tool extension.
Audit Metadata