GH-300 Cheat Sheet: Glossary + Decision Tables
No domain-by-domain walkthrough here — just the terms and the either/or decisions the exam actually tests.
Exam Snapshot
Built from Microsoft's official GH-300 skills outline and GitHub's own Copilot plans documentation, checked September 7, 2026 — not a generic AI-tool glossary with the exam code swapped in.
Copilot Terms Glossary
Every term below maps to a subskill actually listed in the GH-300 skills outline, tagged by which domain it belongs to.
| Term | Domain | What it means |
|---|---|---|
| Agent Mode | Features | Copilot autonomously plans and executes a multi-step coding task across files, running commands and iterating until the goal is met |
| Copilot Edits | Features | A directed multi-file editing session where you specify which files are in scope and review each proposed change |
| MCP (Model Context Protocol) | Features | An open protocol Copilot uses to connect to external tools and data sources beyond the codebase itself |
| Sub-Agents / Agent Sessions | Features | Delegating part of a larger task to a separate agent run, keeping the main session's context window focused |
| Copilot CLI | Features | A terminal-native Copilot client for interactive sessions, script generation, and file management from the command line |
| Coding Agent | Features | An assigned-issue workflow where Copilot works a GitHub issue end-to-end and opens a pull request for review |
| Spaces | Features | A way to group specific files, docs, and context into a reusable bundle for Copilot Chat |
| Spark | Features | Copilot's natural-language-to-full-app scaffolding feature for building small apps from a prompt |
| Content exclusions | Privacy | Repository or organization settings that stop Copilot from using specified files or paths as context |
| Public code filtering | Privacy | An org- or enterprise-level policy that blocks Copilot suggestions matching public code on GitHub above a similarity threshold |
| Proxy filtering | Architecture | A post-processing step that screens a generated suggestion before it reaches the editor |
| Zero-shot prompting | Prompting | Asking Copilot to complete a task from an instruction alone, with no example of the desired output |
| Few-shot prompting | Prompting | Giving Copilot one or more concrete examples (sample I/O, a reference implementation, a test) to demonstrate the desired output |
| Responsible AI (six principles) | Responsible use | Fairness, reliability & safety, privacy & security, inclusiveness, transparency, accountability — Microsoft's standard for evaluating AI system behavior |
| Audit log events | Administration | Organization-level records of Copilot policy and configuration changes, used for governance and compliance review |
Which Copilot Surface Do You Reach For?
Domain 2 (Use GitHub Copilot Features) is worth 25–30% of the exam — the largest single domain — and most of its questions are really this decision, dressed up as a scenario.
| If the scenario says… | Reach for… |
|---|---|
| You want inline suggestions as you type in your editor | IDE inline completions |
| You want to ask a question about code, get an explanation, or discuss an approach | Copilot Chat |
| You need Copilot to plan and execute a multi-step change across several files autonomously | Agent Mode |
| You want to make a scoped set of edits across specific files you choose, reviewing each change | Copilot Edits |
| You want Copilot to connect to an external tool or data source outside the codebase | MCP |
| You want to hand off part of a task to keep the main session focused | Sub-Agents / Agent Sessions |
| You want to generate a script or manage files without leaving the terminal | Copilot CLI |
| You want Copilot to work a full GitHub issue and open a PR without you driving each step | Coding Agent |
Common Mistake: Assuming One Policy Always Wins
Content exclusions and public-code filtering are both configurable at organization level — but neither exists in a vacuum. An enterprise administrator can set an explicit policy that overrides whatever an individual organization would otherwise choose, and an organization's own policy only takes effect where the enterprise hasn't already locked one in. Exam questions that describe a conflict between an org setting and an enterprise setting are testing exactly this hierarchy, not just whether you know the settings exist.
Copilot Plan Tiers
Organization-administration questions (part of Domain 2) assume you know roughly where policy control enters the picture.
| Plan | Best fit | What it adds |
|---|---|---|
| Copilot Free | Individuals trying Copilot | Limited monthly completions, auto model selection |
| Copilot Pro | Individual developers | Full monthly credit allowance, multiple model choice |
| Copilot Pro+ | Power users wanting premium models | Higher credit allowance than Pro, same individual scope |
| Copilot Business | Teams needing central management | Org-level policy controls, per-seat billing |
| Copilot Enterprise | GitHub Enterprise Cloud orgs | Adds enterprise-wide policy override and larger credit pools on top of Business |
Responsible AI: Six Principles
| Principle | One-word anchor |
|---|---|
| Fairness | Equal treatment across groups |
| Reliability and safety | Doesn’t cause harm |
| Privacy and security | Data doesn’t leak |
| Inclusiveness | Empowers more people |
| Transparency | System is understandable |
| Accountability | A person owns the outcome |
GH-300 Quick Answers
What's the actual difference between Agent Mode and Copilot Edits?
Agent Mode plans and executes a multi-step task on its own. Copilot Edits is more directed — you choose the files in scope and review each change before it lands.
Does public code filtering apply the same way to every organization?
No — an org sets its own policy, but an enterprise administrator can lock a policy that overrides it.
Which Copilot plan includes organization-wide policy controls?
Business and Enterprise. Enterprise adds enterprise-wide override capability on top of what Business provides.
Can I print this GH-300 cheat sheet?
Yes — every table here is plain HTML, so a browser print or "print to PDF" renders cleanly.
MSCertQuiz sells practice-exam access for GH-300 and other Microsoft and GitHub certifications; this cheat sheet is written by the same team that builds those questions.
Turning This Into a Passing Score
The domain-by-domain breakdown behind this quick reference.
Scenario questions with full rationale, one story per domain.
Full exam details and the complete practice bank.
A quick diagnostic to see which domain needs the most work.
Tables memorized. Now test the recall.
Start with free GH-300 questions covering every domain above.
Start Free GH-300 Practice