Building better platforms with Azure, GitHub, Terraform and AI
I have been building and using agent skills across Claude Code, Codex and GitHub Copilot, alongside plugins such as Ponytail and Superpowers. I am using Microsoft Agent Package Manager (APM) to manage that collection through a shared manifest. My agent-setup repository holds the configuration and a copy of the global lockfile. The manifest declares what … Read more
Send GitHub Copilot CLI telemetry to Azure Application Insights using an authenticated OpenTelemetry Collector on Azure Container Apps, then query it with KQL.
AI code review becomes noisy quickly on busy pull requests. These are the patterns I would use for managed comments, finding identity, severity, evidence and merge gates.
Update one pull request comment from GitHub Actions instead of creating duplicates after every push, using a stable marker, ownership checks and a simple upsert pattern.
Measuring AI-assisted engineering by activity alone misses the point. Active users, token usage, generated lines of code, and agent sessions are useful signals, but they do not tell you whether the work was reviewable, trusted, safe, or worth the cost.
This post looks at what platform teams should measure instead: workflow success, review effort, guardrail failures, context reuse, cost per useful outcome, and developer confidence.
The Draw.io MCP Diagramming Agent Skill provides a reusable working method for creating clear, editable diagrams with Draw.io MCP. It supports cloud architectures, network topologies, CI/CD pipelines, Kubernetes platforms, event-driven systems, C4, UML and more, while applying consistent guidance for layout, containment, connector routing, labels and validation.
Agent skills are only useful if the agent knows when to use them. A clear description acts as selection metadata, helping the agent load the right guidance, avoid noisy context, and produce more consistent results across repeated engineering tasks.
Agent skills, custom instructions, and MCP configuration are becoming part of the engineering trust boundary. This post walks through using NVIDIA SkillSpector in GitHub Actions to scan AI skill repositories, surface findings in SARIF or PR comments, and make risky agent behaviour visible during normal review.
AI-assisted engineering is moving beyond “can the agent do the task?” and into the same practical concerns platform teams already deal with: cost visibility, ownership, observability, governance, repeatable workflows, and sensible defaults.
This post looks at why AI engineering needs boring platform controls, from APIM policies and token visibility through to Agent Skills, MCP, Terraform modules, OpenTelemetry, and evaluation loops.
AI-assisted engineering is moving beyond proving that agents can complete tasks. The harder question is whether those workflows can be repeated, reviewed safely, kept within sensible cost, and improved over time.
This post looks at why platform teams need to treat AI usage as more than activity metrics, with a focus on token visibility, repeatable agent skills, controlled context, practical guardrails, and measurement that connects AI-assisted work to real engineering outcomes.