audit-agent-background-llm-ops

To enforce inference gateway policies for background LLM ops, this Skill manages the Audit Agent and requires explicit clientTags, budgets, and timeouts for all inference calls. It validates patches via dry-run before applying them to control resource allocation.

When a specific LLM consumer exceeds its CPU budget, this Skill diagnoses the issue using clientTag policies. It allows you to apply surgical corrections to the job pipeline, preventing service degradation and controlling resource allocations.

You can supervise an ollama host and manage job pipelines by deploying the audit-agent through PM2. This Skill implements the Inference Gateway with policies by clientTag to oversee the host supervisor and control CPU and budget allocations.

Yes, you can validate inference gateway patches before applying them. The Skill performs dry-run validation before applying any patches to the ollama host supervisor, ensuring safe policy enforcement and resource management.

Explicit clientTags, budgets, and timeouts are required for all inference calls to control resource allocations in LLM ops. The Skill enforces these inference-gateway policies to manage the background Audit Agent and prevent service degradation.