Skip to content

GH-600 Skills Measured — Full Breakdown

Official source: learn.microsoft.com/credentials/certifications/resources/study-guides/gh-600

Each sub-skill maps to exactly one quest. Click the quest link to start practising that skill immediately.

How to use this page: Work through each domain in order. After reading each sub-skill, rate your confidence (1–5). Return to this page before your exam and ensure every sub-skill is ≥ 4. The Skills Checklist provides a printable version.

Exam Question Archetypes

Understanding the question type is as important as knowing the content. GH-600 uses three primary patterns:

Archetype Description Example stem
Scenario → Diagnosis A failing or misbehaving agent is described — identify root cause "An agent repeatedly requests the same API endpoint. What is the most likely cause?"
Config Selection Choose the YAML/config that satisfies a stated constraint "Which permissions: block grants the minimum privilege for an agent to open a PR?"
Best Practice Select the most appropriate design decision from four plausible options "Which guardrail approach best preserves velocity while preventing irreversible file deletions?"

Keep these archetypes in mind as you read each sub-skill below.

Domain 1 — Prepare Agent Architecture & SDLC Processes (15–20%)

Sub-Skill 1.1 — Integrate agents into the SDLC

Sub-Skill 1.2 — Define boundaries between planning, reasoning, and action

  • Configure agent planning to be distinct from agent execution
  • Configure an agent to output a structured plan
  • Validate agent plans
  • Prevent agent action until the agent checked and approved → Quest: Q2: The Three Sigils — Plan, Reason, Act

Sub-Skill 1.3 — Configure observability and control for autonomous agents

  • Plan and implement the degree of agent autonomy, including guardrails
  • Configure agent to produce inspectable artifacts within standard development tooling
  • Configure human intervention for autonomous agents without slowing delivery → Quest: Q3: The All-Seeing Eye — Observability & Control

Domain 2 — Implement Tool Use & Environment Interaction (20–25%)

Sub-Skill 2.1 — Select and configure agent tools

Sub-Skill 2.2 — Configure MCP servers

  • Add an MCP server as a tool to an agent
  • Configure a GitHub remote MCP server
  • Configure the MCP registries
  • Configure MCP allow lists → Quest: Q5: The MCP Conclave

Sub-Skill 2.3 — Integrate agents within development environments

  • Evaluate the execution context for an agent
  • Configure an agent's scope to a specific repository
  • Configure an agent to be invoked in a CI workflow
  • Configure an agent to use branch-based scope
  • Enable an agent to perform autonomous actions (branches, PRs)
  • Configure an agent to handle environment-specific constraints → Quest: Q6: Bind the Agent to the Realm

Sub-Skill 2.4 — Operate agents with safe execution paths and robust error handling

  • Implement error handling
  • Implement retries
  • Implement rollbacks
  • Implement escalation paths
  • Implement traceability and accountability for agent actions → Quest: Q7: The Shield of Retries

Domain 3 — Manage Memory, State & Execution (10–15%)

Sub-Skill 3.1 — Implement agent memory strategies

Sub-Skill 3.2 — Persist agent state and manage context drift

  • Capture task progress and decisions as durable artifacts
  • Resume agent work without repeating steps or diverging from prior decisions
  • Detect and correct drift during extended agent execution → Quest: Q9: Anchoring the Drifting Agent

Sub-Skill 3.3 — Ensure continuity of agent memory and state across tools and environments

Domain 4 — Perform Evaluation, Error Analysis & Tuning (15–20%)

Sub-Skill 4.1 — Define success criteria and evaluation signals for agent tasks

  • Specify expected outcomes and operational constraints
  • Identify qualitative and quantitative evaluation signals
  • Align evaluation criteria with development intent
  • Generate evaluation signals by using automated scanning tools → Quest: Q11: The Oracle's Rubric

Sub-Skill 4.2 — Analyze agent failures and identify root causes

  • Identify failures using logs, plans, traces, outputs, and workflow artifacts
  • Classify root causes: reasoning errors, tool misuse, context/environment issues → Quest: Q12: The Necromancer's Inquest

Sub-Skill 4.3 — Tune agent behavior based on evaluation results

Domain 5 — Orchestrate Multi-Agent Coordination (15–20%)

Sub-Skill 5.1 — Operate and manage multi-agent workflows

  • Apply an orchestration pattern to coordinate multiple agents
  • Configure agent isolation for parallel execution
  • Detect and resolve agent conflicts (overlapping code changes, duplicated effort, contradictory outputs) → Quest: Q14: The Council of Many — Orchestration Patterns

Sub-Skill 5.2 — Configure observability for multi-agent behavior

Sub-Skill 5.3 — Detect and respond to multi-agent failures and degraded behavior

  • Identify failed, partial, or stalled agent executions
  • Respond to degraded behavior or coordination across agents
  • Implement multi-agent recovery patterns (rollback, human-in-the-loop) → Quest: Q16: When Familiars Fall — Failure & Recovery

Sub-Skill 5.4 — Manage the lifecycle of agents within multi-agent workflows

  • Add agents to existing multi-agent workflows
  • Update, reconfigure, or replace agents without disrupting active workflows
  • Retire agents while preserving auditability and workflow continuity → Quest: Q17: The Agent Pantheon — Lifecycle Management

Domain 6 — Implement Guardrails & Accountability (10–15%)

Sub-Skill 6.1 — Define autonomy levels

  • Classify agent actions by operational, security, and compliance risk
  • Assign autonomy levels to maximize delivery speed while remaining compliant → Quest: Q18: The Autonomy Scales

Sub-Skill 6.2 — Implement guardrails and human-in-the-loop workflows

  • Identify the subset of actions that require human judgment
  • Block actions that violate defined security, compliance, or Responsible AI policies
  • Scope permissions and execution contexts to enforce least-privilege access
  • Require explicit authorization for irreversible or compliance-sensitive changes
  • Preserve execution velocity by minimizing approvals that do not materially reduce risk → Quest: Q19: The Warden's Pact — Guardrails & HITL

Coverage Matrix

GH-600 Sub-Skill Quest Level
1.1 Integrate agents into SDLC Q1 0111
1.2 Plan vs. action boundaries Q2 0111
1.3 Observability & control Q3 1000
2.1 Select & configure tools Q4 1000
2.2 Configure MCP servers Q5 1000
2.3 Dev environment integration Q6 1001
2.4 Safe execution & error handling Q7 1001
3.1 Memory strategies Q8 1001
3.2 State persistence & drift Q9 1010
3.3 State continuity cross-tools Q10 1010
4.1 Success criteria & signals Q11 1010
4.2 Failure root cause analysis Q12 1010
4.3 Behavior tuning Q13 1011
5.1 Multi-agent workflows Q14 1011
5.2 Multi-agent observability Q15 1011
5.3 Multi-agent failure & recovery Q16 1011
5.4 Agent lifecycle management Q17 1100
6.1 Autonomy levels Q18 1100
6.2 Guardrails & HITL Q19 1100
Capstone All 6 domains 1100

Coverage: 19/19 (100%) — no gaps, no duplicates.

Arc hub: Start the full campaign at Agentic Codex — the GH-600 quest arc.

Domain-by-Domain Exam Tips

Domain 1 — What to watch for

Questions tend to be best-practice selection: given a scenario where an agent is "doing too much" or "acting before planning," choose the architecture fix. Know the difference between planning output (structured plan artifact) and action execution — and know what a valid "approval gate" looks like.

Domain 2 — What to watch for

The highest-weight domain. Expect config selection questions (MCP server YAML, permissions: blocks, scope constraints) alongside scenario → diagnosis questions where a CI-triggered agent fails silently. Know the exact structure of an MCP tool definition and the allow-list syntax.

Domain 3 — What to watch for

Questions often appear embedded in Domain 4 or 5 scenarios. The key pattern: an agent "forgets" a prior decision — which memory strategy would have prevented it? Understand the difference between short-term context, long-term store, and external memory, and when each is appropriate.

Domain 4 — What to watch for

Expect scenario → diagnosis: a trace or log excerpt is shown, and you must classify the root cause (reasoning error vs tool misuse vs environment issue). Also expect best-practice questions on how to revise instructions vs tool access vs memory when a particular pattern of failure is shown.

Domain 5 — What to watch for

Orchestration questions test whether you can distinguish sequential, parallel, and hierarchical patterns — and when to use each. Failure-recovery questions ask what signal indicates a stalled sub-agent vs a conflicting sub-agent, and how to respond differently.

Domain 6 — What to watch for

The Autonomy Levels Matrix is heavily tested. Be able to classify an action by risk level and select the correct human-in-the-loop gate. Know what "least-privilege" means in the context of agent permissions and how it differs from traditional RBAC.