Overall readiness
Not started
Add a pack, answer a few questions, and this becomes your honest readiness number.
Your packs
Readiness, coverage and what is due — pack by pack. Set a target date to pace yourself.
Momentum
No sessions yet
Difficulty mix
How your questions are doing right now — unseen questions count as hard.
Drop a .md or .txt exam file here, or .
Multiple files supported
Your exams
Pick up where you left off, or start a timed drill.
No exams yet
Drop a .md or .txt file on the strip above, or add one now.
Keep your quiz library portable
Save exams, answers, flags and categories in one JSON file. Results export is separate.
Loading...
Answer
Why
Exam complete
Here's how the session went.
Analytics
Performance analytics
Insights into your learning progress
Score trend
Last 10 attempts
Domain performance
Question difficulty distribution
Weakest concepts
Tags you have attempted, least mastered first. Click one to drill it.
Illusion of knowing
You answered these wrong but felt sure. Fix these first.
Questions due for review
How to use this dashboard
Everything here works offline; your progress lives in this browser.
One bank, five speeds
Rapid pass (library screen) — the fastest first pass: read the question, say the answer in your head, reveal, be honest (Knew it / Didn't), move on. A whole pack in minutes; it feeds your spacing exactly like a flashcard grade.
Quiz — pick the option. After answering, every wrong option shows why it is wrong right where it sits, and hover any dotted-underlined term (NAT, WORM, RTO…) for its meaning.
Flashcards & Recall — reveal and grade yourself, or type the answer (commands get a mono input).
Trap & Teach — spot the most tempting wrong answer, or explain it in your own words.
Prove — the deep end: operate the question as a system. Only worth spending on questions you keep missing, or the ones with a scene (below).
Prove mode — the Proving Ground
A scene is a question drawn as a live diagram. The Prove button grows a dot when the question on screen has one; Prove the scenes on the library screen starts a session of only those.
1 · Read the system. The question’s path is drawn as a network: the request travels down one rail and the reply comes back up the other. On a decision ladder the stations sit in order; on a timeline they are moments in order. Flip Map / Spine any time — same scene, two shapes.
2 · Probe it (3 checks). Click a station to read that hop’s real evidence — command output, a route table, a log line. You get three, so pick the rows whose state would actually change your mind. Every scene carries a hunt par — the fewest reads a careful call needs (both ends of the failing leg) — and the efficiency line says whether you met it.
3 · Name your suspect. Once a row is read, its small flag is yours: mark the row you believe is the fault (the deciding check, the moment it turned). The app never hints before you commit — your call is your own.
4 · Commit. Answer the question. The reveal then reads your call back against the evidence — you called it, or you named X; the evidence there clears it — the fault is Y — and every piece of evidence gains a verdict tag.
5 · Prove the why. The Reasoning Rail unlocks: place the causal links in order, one card in the pool is a decoy. Then the What-if dial re-routes the scene — flip an assumption and watch the break move. Tell the story walks the scene with the causal links scrolling as captions.
After committing, the audit inverts the skill: every read you gathered sits on the table at once — your one job is to click the piece of evidence that actually decides the case.
Assemble the full chain to earn the proved mark; the Prove the scenes drill puts unproved scenes first. In an exam simulation the flag, rail and dial stay locked until you submit.
Reading logs, diffs and commands
Logs & tickets — click any line to pin it as evidence while you reason; pinned lines highlight and count. Click again to unpin.
Config diffs — the diff starts flat: read both versions and spot the change yourself, then hit Show the changes to check. The green/red was the answer.
Commands — after answering, hover any token of the expected command (--resource-group, Get-Process…) to see its job.
The Case Room
Run a case (dashboard screen) — a whole scenario, not one question: a real estate, real constraints, and a short run of decision stations. Each pick moves the risk / cost / time meters and is graded against the reference plan — no takebacks, exactly like a change window.
One station is a planner board: assign every workload a migration strategy (the 7 Rs) and a wave, then lock the plan. Dependencies are flagged on each row — a workload cannot reasonably leave before what it depends on.
The debrief lays your plan beside the reference run, decision by decision, and the run feeds concept mastery for the tags the case carries. Esc leaves a case at any time.
The Architect Workbench
Architect workbench (dashboard screen) — six guided architectures you build, break and repair: two AWS landing zones, two Azure landing zones, a grounded AI assistant and a tool-using agent. Each one is a diagram that starts almost empty and fills in as you decide.
1 · Read the diagram. Select any box or arrow — with the mouse, or with Tab and Enter — to read its purpose, what it depends on, where the trust boundary sits and how it should be configured.
2 · Decide. Every decision is consequential: the components you choose are drawn, the ones you skip never appear, and the cost / security / resilience chips move. There is a reference design behind the score, and the tempting answer is usually a trade.
3 · Repair. When a stage’s decisions are in, a fault is injected against your design — read the evidence, then pick the repair. A stage only unlocks when the fault is repaired.
4 · Trace it. Play, pause, step and reset a simulated activity trace over your own design — the simulated-vs-real line is stated on the panel. Then the assessment: sequence the build order and point at the evidence that proves the fault. No multiple choice.
A scenario’s result feeds concept mastery for the pack it belongs to. Everything is deterministic and offline: fictional organisations, no cloud calls, no credentials, no model calls, and indicators are illustrative rather than prices or measurements.
Concepts & twins
Under every explanation sit concept chips with a count — the other questions that teach the same idea. Click one to see them all, or drill just that concept.
A dashed Confusable with chip offers the classic exam confusion side by side — same concepts, different verdicts. Reading the distinction once is worth ten repeats.
Keyboard
| A–E | pick an option (Quiz / Trap / Prove) |
| Space | reveal flashcard · reveal rapid pass · flag the question |
| 1–4 | grade a flashcard |
| 1–9 / F | Prove: read that row / name the last one you read |
| 1 / 2 | Rapid pass: knew it / didn't |
| ← → | previous / next question |
| Esc | close a dialog, step back |
| Ctrl+K | command palette |
Where your progress lives
Answers, mastery, notes, collections and every proved scene are stored in this browser only — nothing is uploaded. Use Backup in the profile menu to export a file you can restore on another machine; backups merge, so restoring never loses newer work.
Question format guide
Create exam files with .md or .txt format
Quick start
Every exam file needs:
- # Exam Title — file title (one # hash)
- ## Module Name — domain/topic (two ## hashes)
- ### Q1, Q2, Q3... — questions (three ### hashes)
- 4 options (A, B, C, D) — exactly 4 choices
- Correct answer: B — which option is right
Markdown example
# Azure Fundamentals Exam ## Cloud Concepts ### Q1 What is cloud computing? - **A.** Using physical servers - **B.** Internet-based services - **C.** Local storage only - **D.** Install software locally **Correct answer:** B **Explanation:** Cloud computing delivers computing services over the internet.
Text (.txt) format
# Python Quiz ## Data Types ### Q1 Which is immutable? - A. List - B. Dictionary - C. Tuple - D. Set **Answer:** C **Explanation:** Tuples cannot be modified after creation.
Accepted formats
- - **A.** text
- - A. text
- A) text
- **Correct answer:** B
- **Answer:** B
- Answer: B
Optional: types, tags & drills
These extras are optional. Files without them keep working exactly as before.
### Q7 **Type:** recall **Tags:** azure.identity, managed-identity **Objective:** Managed identities in Entra ID **Ref:** https://learn.microsoft.com/entra/identity/managed-identities-azure-resources/overview What lets a VM reach Key Vault without storing a credential? **Accepted:** managed identity ;; managed identities ;; MSI **Explanation:** A managed identity is an Entra ID identity bound to the resource. **Trap:** Service principal — same idea, but you must manage the secret. ### Q8 **Type:** command **Tags:** powershell.az.vm List every VM in resource group prod-rg. ```powershell Get-AzVM -ResourceGroupName prod-rg ``` **Accepted:** Get-AzVM -ResourceGroup prod-rg
- Type: mcq | recall | command | blank | scenario
- Tags: comma-separated concept tags
- Accepted: alternate answers, split by ;;
- Trap: the tempting wrong answer
- Why A is wrong: per-option rationale
- Hint: shown before you answer
- Ref: / Objective: study links
- Scenario: + Next: chain steps
Important rules
- Use .md or .txt
- 2+ options per MCQ
- Plain text only
- Add explanations
- HTML tags
- Fewer than 4 options
- .docx or .pdf files
- Missing answers
Example files
Download these ready-to-use example files:
Need detailed help? Check TUTORIAL.md in the repo
Batch upload exams
Select multiple .md or .txt files to upload at once
Results
Exam
New question
Proving Ground scene (optional)
Turn the question into an operable system: hops, evidence, the drop leg, the causal chain, then the what-if states. The builder writes the directives and proves them with the app's own parser.
This markdown is generated by the app's own parser, so it round-trips. Drop it back into packs/ and run npm run build:packs.
Content coverage
Where your bank is thin, measured against the official exam weightings.