self-healing-orchestrator

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill explicitly integrates Confidential and Personal knowledge tiers and "prioritizes the most specific secrets" for automated operational actions, which implies the agent will access and use secret values and creates a realistic risk those secrets could be embedded verbatim in generated commands/requests unless explicitly constrained to use secure patterns.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill explicitly instructs the agent to restart services, apply patches/rollbacks, and deploy hotfixes—actions that modify system state and likely require elevated privileges—so it directs the agent to change and potentially compromise the machine.