In the world of GitOps, a successful sync doesn't always mean a healthy application. Standard static analysis tools like kube-linter or kubeconform are excellent at catching syntax errors, but they often produce fragmented, noisy data that leaves developers guessing at the "why" and "how to fix".
This session introduces LLM-powered skills. Skills provide focused context and real-time retrieval without external dependencies, making them ideal for secure, offline infrastructure environments.

Using a Manifest-Validator plugin, we demonstrate how to equip ArgoCD with specialized, modular abilities that go beyond simple text generation. We will deep-dive into:
1) The Sidecar Architecture: How we consolidated linter outputs and model configurations into app-specific ConfigMaps to create a self-contained validation engine
2) Actionable Remediation: How the generated diagnostic reports can help provide developers with root cause analysis
3) The Live Audit: How the AI skills inspects live cluster resources to suggest security hardening and cost optimization opportunities based on production best practices

Attendees will leave with a blueprint for building agentic DevOps tools that don't just find problems, but possess the specialized expertise to solve them.

90% of GenAI systems fail in production because teams struggle to evaluate and resolve failures. Observability tools like OpenTelemetry and MLflow log massive traces, but hide critical reasoning failures. They tell you where a request traveled, but not why an LLM agent lost its way. This session moves beyond basic observability into the realm of failure mode analysis for agentic systems. Using a real world AI system as a case study, we will dissect why traditional monitoring fails to catch silent agent failures like context window saturation and tool-calling loops.

Attendees will walk away with a practical toolkit for:
1) Defining a Failure Taxonomy: Moving from vague bad outputs to structured categories like retrieval gaps or reasoning drift
2) Building a Tracing-to-Resolution Pipeline: Leveraging OpenTelemetry and MLflow traces to map anomalies directly to code-level fixes, eliminating the guesswork of manual prompt engineering
3) Human-in-the-Loop Evaluations: Scaling quality control by using LLMs to surface high-risk anomalies for critical human review

Learn to stop operating blind and build resilient, production-grade AI systems that your teams can actually trust.