online-evals

Pass

Audited by Gen Agent Trust Hub on Jul 28, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXFILTRATION]: Interacts with LaunchDarkly's official API at app.launchdarkly.com to manage judge configurations, retrieve targeting data, and attach evaluators to variations.
  • [EXTERNAL_DOWNLOADS]: References and utilizes official LaunchDarkly SDKs (ldai, ldclient) and technical documentation from the vendor's public GitHub repositories and official domains.
  • [PROMPT_INJECTION]: Allows users to define custom system prompts for LLM-based evaluation judges. This is the primary intended functionality of the skill and allows for domain-specific scoring criteria within the user's LaunchDarkly project environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 28, 2026, 12:52 PM
Security Audit — agent-trust-hub — online-evals