Final Exam Attack Plan | NetForge Cloud Learning

[!!]

Around 57–60 questions, 120 minutes, mix of single-choice and multi-select scenario questions. Passing threshold around 70%.

[!!!]

Practical pass target: aim for 80% in practice so you have a safety margin on exam day.

[TIME]

Pace model for 57 questions / 120 minutes:

Average budget: ~2 minutes 6 seconds per question
First pass target: finish all questions by minute 95–100
Final review buffer: 20–25 minutes for flagged items

[TRAP]

Spending 4–5 minutes on one uncertain question early in the exam.

[ACTION]

Hard rule: if not solved by 2 minutes and you are between two options — flag and move on.

[!!!]

Priority order by weight:

D4 Effective Use

45%

D2 Features & Plans

25%

D3 Setup & Config

20%

D1 Responsible AI

10%

[ACTION]

Study allocation if time is limited:

45% of effort on Domain 4 drills
25% on plan/feature boundaries
20% on setup/admin/exclusions
10% on responsible AI fundamentals

[TRAP]

Equal study time across all domains — the exam is not weighted equally.

3.1 Enterprise-Only Trio

[!!!]

Memorise exactly: Knowledge bases, PR summaries, Bing search in Chat

[CHECK]

"If the scenario requires any of the three enterprise-only capabilities, the correct plan is Enterprise."

[TRAP]

"Business includes knowledge bases." — False.

3.2 Slash Commands Core Map

[!!!]

/explain → understand code behaviour
/fix → correct broken or failing code
/tests → generate unit tests
/doc → generate documentation comments
/new → scaffold new file or component

[!!]

Also know: /clear resets chat thread; /help shows available commands.

[TRAP]

/explain and /fix are interchangeable — False. Explain describes; Fix changes.

3.3 Context Variable Core Map

[!!!]

@workspace → whole local project context
@file → one specific file context
@terminal → terminal errors / output
@github → GitHub platform context (issues, PRs, repo)

[!!]

Attachments:

#file → include a named file
#selection → include currently selected code

[TRAP]

@workspace means internet search — False. It is local project only.

3.4 Plan Boundary Snapshot

[!!!]

Individual: personal developer use, no organisation controls
Business: org controls — seats, policy, exclusions, audit
Enterprise: Business + enterprise-only trio + deep governance

[TRAP]

Individual can manage organisation policies — False.

4.1 Generative AI Mechanics

[!!!]

Exam-safe phrasing: "Copilot predicts likely next tokens using context and learned patterns; human verification is always required."

4.2 Training vs Runtime Context

[!!]

Training: largely from public code and text corpora
Runtime: uses current context — open files, cursor, project signals

[TRAP]

"Copilot is retrained live on your private repository during your session." — False.

4.3 Privacy and Data Handling

[CHECK]

If a question asks for strict org-level governance and non-retention style controls, which plan family? → Business or Enterprise.

4.4 Security Risk Patterns

[!!!]

Never trust generated code blindly in:

Authentication / authorisation logic
Secret handling
SQL query construction
Input validation
Cryptography

[ACTION]

Mental security checklist for any generated snippet:

Are secrets hardcoded?
Are untrusted inputs validated/sanitised?
Are queries parameterised?
Are dependencies current/safe?
Are error paths secure and explicit?

[TRAP]

"Because Copilot's suggestion compiles, it is secure." — False.

4.5 Bias and Hallucination

[CHECK]

Bias = skewed suggestions reflecting historical coding patterns. Hallucination = invented APIs, functions, or packages that look real but don't exist.

5.1 Plan Selection Decision Tree

[MAP] Plan-choice flow:

Does the requirement mention knowledge bases, PR summaries, or Bing chat search?
→ Yes: Enterprise · No: continue
Does it require org policy control, seats, exclusions, audit, SSO?
→ Yes: Business (or Enterprise) · No: continue
Is this a single developer with no central governance requirement?
→ Yes: Individual

[TRAP]

"50 developers means Enterprise by default." — Not always. Scale alone does not force Enterprise; enterprise-only features must be required.

5.2 Completion vs Chat vs Agent

[!!!]

Inline completion: passive, in-flow suggestions as you type
Chat: active question/answer and generation flow
Agent mode: autonomous multi-step execution

Question clue words: "as they type" → completion · "ask for explanation/fix/tests" → chat · "multi-file autonomous" → agent

5.3 Interfaces and Surfaces

[!!]

IDE chat panel, inline chat, GitHub.com chat, CLI assist, mobile awareness at high level.

[TRAP]

Over-claiming parity — not every interface supports every advanced capability identically.

6.1 IDE and Environment

[!!]

VS Code is the fullest common reference point for exam scenarios. Neovim is commonly tested as limited compared to full IDE chat feature sets.

6.2 Admin Controls and Governance

[!!!]

Admin-level controls frequently tested:

Seat assignment / management
Policy controls
Audit / log visibility
Feature toggles per org policy posture

[CHECK]

If a question asks who can enforce policy across many repos in an org, the answer should point to → organisation admin role level.

6.3 Content Exclusions (Critical)

[!!!]

What exclusions DO: Prevent selected content from being used as Copilot context.

What exclusions do NOT do:

They do NOT block file access for users
They do NOT disable Copilot globally

[MAP] Exclusion reasoning:

Need to stop AI context usage of sensitive files? → Content exclusions
Need to stop humans reading files? → Repository permissions, not exclusions

[TRAP]

Confusing context exclusion with repository permission / access control.

7.1 Prompt Engineering Structure

[!!!]

Four-part prompt anatomy:

Context
Task
Constraints
Output format

Ordering rule: Give context before asking for task execution.

[CHECK]

Build this prompt in one sentence:

Context: Python API endpoint with failing auth tests
Task: fix token validation
Constraints: do not add new dependencies
Format: unified diff

7.2 Command and Context Pairing

[MAP] Intent → tool:

Understand code → /explain with #selection or @file
Fix runtime / test error → /fix with @terminal + file context
Generate tests → /tests with #file or #selection
Add docs → /doc with #selection

[TRAP]

Using broad @workspace when narrow file/selection context would be more precise.

7.3 Agent Mode vs Standard Chat

[!!]

Agent mode is strongest when: multi-file edits, multi-step sequences, or some autonomous orchestration is acceptable.

Standard chat is strongest when: you want explicit step-by-step review and tighter manual control.

[TRAP]

Choosing autonomous mode for a tiny, one-line clarification task.

7.4 Practical Developer Workflows

[!!]

Exam-aligned use cases:

TDD support: generate tests, then refine implementation
Rubber-duck debugging: explain → hypothesise → fix
Documentation uplift with /doc
Scaffold / refactor loops with clear constraints

[MAP] SCAN method — use on every scenario question:

S = Scope — single file, project-wide, terminal, GitHub platform
C = Capability — explain, fix, test, document, administer, choose plan
A = Authority — individual dev, org admin, enterprise admin
N = Non-negotiables — enterprise-only features, exclusions semantics, privacy posture

[CHECK]

If you cannot identify all four SCAN parts, re-read the question stem before answering.

[!!!]

Trap cluster A — Overclaiming AI capability:

"Automatically correct and secure without review"
"Guaranteed correct output"

[!!!]

Trap cluster B — Scope confusion:

@workspace vs @github
Exclusion vs access control

[!!]

Trap cluster C — Boundary confusion:

Business vs Enterprise features
/fix vs /explain intent mismatch

[TRAP]

Watch out for answer options containing:

"always"
"guarantees"
"all cases"
"no review required"

[TIME]

Three-pass strategy:

Pass 1 (0–70 min): Solve all straightforward items quickly. Flag uncertain questions — do not stall.
Pass 2 (70–100 min): Return to medium-difficulty flagged items. Use elimination + SCAN framework.
Pass 3 (100–120 min): Final review of hard flags. Validate multi-selects carefully. Ensure no unanswered items.

[CHECK]

At minute 60 you should be approximately half-to-two-thirds complete.

[TRAP]

Changing answers without a clear reason. Change only when new evidence from the stem or constraints is identified.

Day	Focus	Actions
Day 7	Baseline & Gap Map	25–30 question timed mini-mock. Categorise misses by: PLAN, COMMAND, CONTEXT, EXCLUSION, RESPONSIBLE_AI, SETUP. Build your top-3 weakness list.
Day 6	Plans & Features Lock-in	Drill 40 plan-selection and feature-boundary prompts. Flashcards for enterprise-only trio and common distractors.
Day 5	Commands & Context Mastery	Drill intent → slash command mapping. Drill scenario → context variable mapping. Target: 90%+ accuracy.
Day 4	Setup / Admin / Exclusions	Governance, auth, IDE support edges, exclusions behaviour. Write one-sentence definitions from memory.
Day 3	Domain 4 Intensive	Prompt engineering reps: context-task-constraints-format. Mixed scenario sets emphasising effective-use decisions.
Day 2	Mixed High-Pressure Mock	Full-length or near-full timed mock. Review only mistakes and uncertainty points. Simulate exact pace with strict flagging rules.
Day 1	Light Consolidation	No heavy new content. Read all [!!!] and [TRAP]. One short confidence set (10–15 questions). Sleep well.

[ACTION]

T-24 to T-12: Read sections 3, 5, 6, 7, 9, 10 of this plan. Rehearse plan and command maps aloud.

T-12 to T-2: Very light review only. Stop difficult new question sets. Protect focus and rest.

T-60 minutes: Re-read the non-negotiable memory set. Write a quick brain-dump at exam start:

Enterprise-only trio
Slash command map
Context variable map

"I will identify scope, capability, authority, and non-negotiables for every scenario. I will eliminate exaggerated claims, choose precise technical truth, and protect my time budget. I do not need perfect certainty on every item; I need disciplined decisions across the full exam."

Score each item 0–2 · 0 = weak · 1 = partial · 2 = confident without notes

#	Skill	Your Score (0–2)
1	Enterprise-only trio recall
2	Plan decision tree accuracy
3	/explain vs /fix distinction
4	/tests and /doc usage boundaries
5	@workspace vs @github distinction
6	@terminal usage scenarios
7	Content exclusions semantics
8	Responsible AI core risks
9	Agent vs standard workflow judgement
10	Time-management discipline

Score interpretation:

18–20 → Exam ready ✓
15–17 → One focused revision block needed
12–14 → Two targeted blocks needed
0–11 → Rebuild core foundations before sitting exam

Use these for your final review pass before exam day.

[!!!] Card 1 — Enterprise-only?

Knowledge bases, PR summaries, Bing search in Chat

[!!!] Card 2 — Fix vs Explain

/fix changes code. /explain describes code.

[!!!] Card 3 — Scope anchors

@workspace local project
@github platform context
@terminal console output

[!!!] Card 4 — Exclusions

Exclude AI context usage — NOT user access permissions.

[!!] Card 5 — Prompt anatomy

Context → Task → Constraints → Output format