Ultimate Agentic AI DevOps with Claude Code

Explore how cloudA.md acts as the AI onboarding document at the project root, loaded automatically, differentiating from readme.md while detailing architecture, conventions, tools, and rules for cloudA code sessions.

This lecture shows you write your first cloudA.md for a static HTML/CSS site deployed on AWS using S3 and CloudFront, covering project overview, architecture, commands, conventions, and safety.

Explore how a well written CloudAid.md guides end-to-end devops workflows, replacing hours of manual work with a single document that defines architecture, conventions, safety layers, and automated deployment.

Discover how cloud.md loads context into the agent before prompts, shaping decisions, tool usage, and code generation within the cloud core architecture (ide, agent, tools, llm).

Learn how skills transform a generic llm into specialist roles like a terraform expert. Explore built-in and custom skills and how the agent uses cloud.md to guide the llm.

Cloud.md auto loads to provide project context, conventions, architecture, and role, always shaping every llm response, while skills are on-demand instructions triggered by slash tf learn.

Explore why on demand skills make DevOps engineers reliable by providing consistent Terraform plan guidance, safety controls, and repeatable workflows for scalable infrastructure.

Explore how skills are defined as markdown files with a yaml formatter and body, stored in .cloud skills, and loaded through progressive disclosure via manual slash commands or description matches.

Discover how to deploy a site on AWS using S3 and CloudFront by building four skills: scaffold terraform, tfplan, tfapply, and deploy to generate infrastructure, plan, apply, and publish.

Scaffold terraform to generate a complete S3 and cloud front static site infrastructure. Use region and project name arguments and a template-spec, then run tf plan before applying.

Walk through the tfapply and deploy skills to run terraform apply and deploy site files to S3, invalidate CloudFront, and verify the distribution status.

Scale a live site with four agentic skills: scaffold, plan, apply, deploy, using terraform. Generates S3 and cloud front resources, and handles plan, apply, and cache invalidation.

Explore sub-agents that prevent context pollution by assigning read-only security audits, write-enabled terraform tasks, and read-only cost checks to isolated, fresh-context workers.

Compare skills and sub-agents to decide whether to load instructions into the main chat or run an isolated task; use skills for context, sub-agents for standalone tasks like auditing.

Define sub-agents using the seven-field formatter, ensuring name matches the file and description outlines when to use the agent, then manage tools, memory, and model choices.

Explore three specialized sub-agents: security auditor, cost optimizer, and TF writer that audit Terraform infrastructure, optimize costs, and generate Terraform code via the MCP server.

Run three sub-agents on live terraform to audit security and optimize costs, orchestrate hands-off tasks, and apply fixes with the TF writer agent for automated remediation.

Explore how MCP, an open standard called the Model Context Protocol, enables live data. Use Terraform MCP and AWS MCP to get direct, structured answers from real sources.

Explore how MCP architecture connects the cloud code agent with Terraform and AWS MCP servers via the .mcp.json config, delivering tools, resources, and prompts for real-time data.

Learn MCP transport and scopes with studio as the local transport, and project and user scopes to share configurations, enabling fast, consistent cloud code setup across the team.

Place .mcp.json at your project root and deploy terraform and aws mcp servers with docker to fetch live documentation. Manage credentials in settings.local.json and use $FAS syntax for variable expansion.

Explore settings.local.json walkthrough, configuring environment variables, AWS credentials, and region, enabling MCP servers for Terraform and AWS, and managing personal permissions while keeping credentials private.

Explore popular MCP servers, from AWS and Terraform to Notion and Slack, learn how to install and manage MCP servers with Cloud Code, including transport options and scopes.

Connect Terraform and AWS mcp servers in cloud code, confirming docker, python with uvx, and valid credentials for real-time data access.

Demonstrates live ci/cd setup with AWS OIDC authentication, creates an IAM role and Terraform workflow, and runs infra-audit and infra-status to deliver a unified health dashboard.

Apply everything learned in the course to the capstone by deploying a React app to AWS with Cloud Code, featuring a build step, S3 sync, and CloudFront with OIDC.