CertifyDeckCargando contenido…
The official Microsoft Certified: Agentic DevOps Engineer Expert syllabus, organized by domain. Check off each point as you review it.
Inicia sesión para guardar tu progreso entre sesiones.
Design traceability and flow of work
Design the flow of work by connecting phases: plan in Azure Boards or GitHub Projects (columns, split columns, WIP limits, Definition of Done), build with automatic linking of work items to builds, deploy through environments that record history and deployed work items, and operate by creating work items from Application Insights with their telemetry context.
Create and target Azure DevOps environments ↗Configure pipelines to support work tracking ↗
Feedback cycles in Azure DevOps rely on subscriptions: personal, team, project or global, delivered by email or through service hooks to external services such as Teams, Service Bus or Datadog. Delivery by role limits alerts to the people involved and Skip initiator avoids notifying whoever made the change. In GitHub Actions, scheduled workflow notifications follow the creator or whoever re-enables it.
Choose the process to fit the team: Basic (epics, issues, tasks), Agile (user stories), Scrum (backlog items with Approved and Committed states) or CMMI (change control, risks and reviews). Organize work with area paths per team or component, which can also restrict access, and iteration paths per sprint, and use delivery plans to see several teams and their dependencies on a calendar.
Default processes and process templates ↗Use team delivery plans in Azure Boards ↗
Traceability connects work items with branches, commits, pull requests, builds, releases and tests. In Azure Repos, the Check for linked work items policy enforces links; with GitHub, AB# in commits or descriptions links items and, with fixes, transitions them when merged into the default branch. Requirement-based test suites and traceability reports measure the quality of each requirement.
Link GitHub commits, pull requests, branches, and issues to work items in Azure Boards ↗About end-to-end traceability ↗
Design and implement success metrics for DevOps
Turn each KPI into a measurable metric: lead time (creation to Completed) for customer commitments, cycle time and the cumulative flow diagram for efficiency and bottlenecks, pass rate, duration and test failure reports for delivery quality, and GitHub security overview (Coverage, Remediation) for security, using the audit log when compliance needs consistent data.
Configure lead time and cycle time widgets ↗Security overview ↗
With the Azure DevOps MCP Server in agent mode, an assistant turns natural language questions into WIQL queries and tool calls over work items, builds and tests. Analytics exposes OData (LeadTimeDays and CycleTimeDays only for completed items) for Power BI. Azure Copilot generates sample KQL for Azure Monitor logs, and Copilot usage metrics measure AI adoption.
Enable AI assistance with Azure DevOps MCP Server ↗Lead time and cycle time sample report ↗
Configure collaboration and communication
Document the project with Azure DevOps wikis (a project wiki or code published as a wiki, kept in sync with the branch), a README in .github, root or docs, release notes generated and configured in .github/release.yml, a LICENSE file, and CODEOWNERS, where the last matching pattern wins. Mermaid draws process diagrams, and Copilot drafts pull request summaries starting from a blank description.
Set requirements for agents with instructions in .github/copilot-instructions.md, .instructions.md files with applyTo, and AGENTS.md; skills in .github/skills with SKILL.md; and custom agents whose tools field limits tools. Copilot cloud agent pushes only to its copilot/ branch and cannot approve its pull requests. Scheduled agentic workflows run with read permissions and write only through safe-outputs.
Adding repository custom instructions for GitHub Copilot ↗GitHub Agentic Workflows ↗
Azure DevOps MCP Server can be remote (https://mcp.dev.azure.com/{organization}, HTTP and Entra ID), filtered with X-MCP-Toolsets, X-MCP-Tools or X-MCP-Readonly, or local (stdio, PAT or Entra ID). The remote GitHub MCP server uses OAuth or a PAT. For Copilot cloud agent, the repository JSON supports only tools, uses COPILOT_MCP_ secrets, and the built-in GitHub server reads only the current repository.
Set up the remote Azure DevOps MCP Server ↗Configure MCP servers for your repository ↗
GitHub webhooks and Azure DevOps service hooks push events (pushes, reviews, work items) to external systems without polling. MCP servers give Copilot tools: for Copilot cloud agent they are configured per repository and read Agents secrets prefixed COPILOT_MCP_. Agent skills are folders with a SKILL.md file (name and description) under .github/skills.
Configure MCP servers for your repository ↗Integrate with Service Hooks - Azure DevOps ↗
The Azure DevOps MCP Server gives AI assistants in agent mode access to work items, pull requests and pipelines. The remote server (https://mcp.dev.azure.com/{org}) requires Microsoft Entra; the local one runs with npx over stdio. X-MCP-Toolsets and X-MCP-Readonly headers limit its tools, and Azure Boards can start Copilot on GitHub repositories.
Set up the remote Azure DevOps MCP Server ↗Use GitHub Copilot with Azure Boards ↗