GH-300 study plan

Your step-by-step plan

1. Map the blueprint and book a date

   Day 1

   Read the six domains and their weights. Book a provisional date now: a fixed date turns open-ended study into a plan and is the strongest predictor of actually sitting. Note that Using Copilot features at 28 percent and Responsible use at 18 percent are nearly half the exam between them, so plan the heaviest study there, with productivity, privacy, prompting, and architecture filling the rest.

2. Build the surface and control decision trees

   Week 1

   Before drilling any domain, build the maps the whole exam rests on. Fix the six surfaces by what each does (inline, Copilot Chat, Edit Mode, Agent Mode, Plan Mode, GitHub Copilot CLI) and the controls by their real job (content exclusions, duplication detection, organisation Copilot policy, custom instructions, Copilot Spaces, Model Context Protocol). Use the recall prompts in this guide: cover the answer, choose the surface or control from the need, then reveal. If you cannot pick from the need alone, you do not own it yet.

3. Go deep on features and responsible use

   Weeks 1 to 2

   These two domains are nearly half the exam, so they get the most time. Drill the surface split until choosing Chat for a pure explanation and the CLI for a terminal command is automatic, and learn what each plan tier and the organisation policy unlock. In parallel, make the responsible-AI rule a reflex: confidence is not correctness, the accountable developer validates output, and recent external facts must be confirmed against the authoritative source. Read the worked explanation on every practice question, including the ones you got right.

4. Lock productivity and privacy controls

   Weeks 2 to 3

   Productivity and the privacy controls are dependable marks once drilled as need-to-mechanism matches. Fix code generation from a descriptive signature and comment, the custom instructions file for persistent test conventions, and the GitHub Copilot CLI for terminal commands. For privacy, separate content exclusions (withhold named files as context) from duplication detection (suppress public-code matches), and learn the limits: neither is a licence guarantee, and a hardcoded secret in a suggestion is untrusted output to replace.

5. Cover prompting and the data and architecture model

   Week 3

   Drill the prompt structure (goal, then context and constraints, then expected output), the zero-shot versus few-shot decision, and the two properties of effective few-shot examples. Trace one inline completion end to end through the Copilot proxy and learn what the filtering pipeline does to a candidate, plus the core LLM limitation that a model only uses training data or supplied context. Reject the distractors that overstate the data flow or invent request-time training.

6. Drill weak domains, then space the review

   Week 4

   Use your per-domain accuracy to attack the domains dragging you down, not to re-read what you already know. Then space it: revisit each domain's recall prompts after a few days and again later in the week. Spacing roughly doubles what sticks compared with cramming, and it is the cheapest gain available before the exam.

7. Sit a timed mock and calibrate

   Weeks 4 to 5

   Take at least one full timed mock under exam conditions to rehearse pacing and the flag-and-return habit across the whole question set in the time allowed. Treat the score as a per-domain readiness signal, not a single number, and review every missed question - naming the need you misread and the feature you confused - before you book or sit.

Frequently asked questions