README file from
Github🧠 CertFS Quiz
An Obsidian plugin. It quizzes you on a note, a selection, several notes or a folder, in exam style. It also re-quizzes your errors with new versions until you master them.
- 🤖 Questions come from an OpenRouter model (free ones work).
- 📚 Subjects are set in Settings: a folder + an optional question style. Nothing is hardcoded.
- 🏃 Presets included: CSM (Scrum Guide 2020), PMP (PMI mindset, "what FIRST?"), CPMAI.
- ❌ Wrong answers → error notes in that subject's errors folder.
- 📊 Every answer goes to a local history → stats + no repeated questions.
- 🔁 Shareable: copy the folder (without
data.jsonandhistory.json) to any vault.
📑 Contents
- 📥 Install
- ⚙️ Setup (once)
- 🎯 Daily use
- 🧩 Question types
- 🗂️ How it picks the subject
- ❌ Error notes (automatic)
- 🔁 Errors → Archive (mastery)
- 📊 Answer history + stats
- 💸 Free limits (OpenRouter)
- 🛠️ Errors and fixes
- 🔒 Privacy
- 📝 What's new
- 🧑💻 For developers
📥 Install
- 🛍️ Community plugins: Settings → Community plugins → Browse → search CertFS Quiz → Install → Enable.
- 🧪 Before it is listed (BRAT): install the BRAT plugin → Add beta plugin →
KaramGuliyev/CertFS. - 📦 Manual: download
main.js,manifest.json,styles.cssfrom the latest release into<vault>/.obsidian/plugins/certfs-quiz/→ restart Obsidian → enable CertFS Quiz.
⚙️ Setup (once)
- 🔑 Make a key at openrouter.ai → Keys.
- 📋 Settings → CertFS Quiz → paste the key.
- 📥 Click Load free models → pick one that ends in
:free.- 🧹 The list shows only models that can chat (no embed, TTS, rerank).
- 💲 On openrouter.ai, leave the key's credit limit empty. Don't set it to $0.
- 📚 Settings → Subjects: check the list. On first run, presets whose folder exists are added.
- ➕ Add subject → name, folder, rules note (optional), question style (optional).
- 🔍 Folder and Rules note boxes search the vault as you type → click a result.
- 📦 Add presets → adds CSM, PMP, CPMAI (edit their folders to match your vault).
🎯 Daily use
| Do this | How |
|---|---|
| 📖 Quiz the open note | 🧠 ribbon icon, or Ctrl+P → Quiz this note |
| ✂️ Quiz part of a note | Select text → Ctrl+P → Quiz selected text |
| ⚙️ Pick settings | Count (1–15), difficulty, optional focus, question types |
| 🚀 Start | Generate |
| 📚 Quiz several notes | Ctrl+P → Quiz multiple notes → search + tick notes, or select notes in the file explorer (Ctrl/Cmd+click) → right-click → CertFS Quiz: quiz N notes |
| 📁 Quiz a whole folder | Ctrl+P → Quiz a folder → pick it |
| 🔁 Re-quiz errors | ↺ ribbon icon, or Ctrl+P → Quiz my errors |
| 📦 Test the archive | Ctrl+P → Test my archive |
Folder quiz:
- 📚 Uses every note in the folder and its subfolders.
- 🚫 Skips
Errors/,Archive/,Mocks/and files starting with_(Settings: Folder quiz skips). - ⚖️ Each note gets a fair share of the 24,000-character limit.
Multiple notes quiz:
- ✅ The open note is ticked to start with; search by name or path, tick more.
- 📚 Same flow as a folder quiz: one request, questions spread across the notes.
- 🗂️ Subject = the one most picked notes belong to (errors go there).
In the quiz:
- 💡 Hint → a small push, not the answer.
- 🤷 Not sure → counts as a miss, so it gets reviewed.
- ✅ Check → right/wrong, the correct answer, and why.
- ✋ I was right → short answer graded wrong but you know it's right. Counts as correct, no error note.
- ➡️ Next → the next question.
At the end:
- 📊 Score % and the topics you missed.
- 👀 Review missed → each missed question with the right answer and why. No re-test.
- 🆕 New questions on missed topics → fresh questions, no repeats.
🧩 Question types
| Type | What you do |
|---|---|
| 🔘 Multiple choice | Pick 1 of 4 |
| 🎭 Scenario | A situation → pick the best action (1 of 4) |
| ☑️ Multi-select | Pick all correct options (exact set) |
| ⚖️ True / False | Pick one |
| ✍️ Short answer | Type it. The AI grades the idea, not the wording |
| 🔗 Matching | Match each term to its pair |
- 🎛️ Turn types on/off in Settings or in the quiz setup.
- 🔀 Options are shuffled every time.
🗂️ How it picks the subject
- 📁 The note's path is matched against each subject's folder.
- 🎯 Nested subjects: the deepest folder wins.
- 🧩 No match → General style, and mistakes are not saved.
| Subject setting | Used for |
|---|---|
| Name | exam: in error notes, the subject dropdown, stats |
| Folder | Which notes belong to it; errors go to <folder>/Errors/ |
| Rules note | **Rule:** → [[...]] line in error notes (empty = no line) |
| Question style | Optional. Empty → the AI guesses the exam from name + folder. Fill it to steer wording, traps, scenario type |
Presets (default folders, edit them in Settings):
| Folder | Style | Rules link |
|---|---|---|
1 CSM/ |
CSM (Scrum Guide 2020) | Scrum Rules |
2 PMP/ |
PMP (PMI mindset) | Mindset Rules |
3 CPMAI/ |
CPMAI | CPMAI Rules |
❌ Error notes (automatic)
- 📄 File:
<subject folder>/Errors/YYYY-MM-DD <topic>.md. - 🏷️
ErrorsandArchivenames can be changed in Settings (old notes are not moved). - 🧾 Same frontmatter as
99 Templates/Error.md. - 🏷️
source: "CertFS Quiz (<note>)",status: new,next_review= tomorrow. - 🏠 So it shows up in Home under errors to redo.
- 🔁 Same topic, same day, different question → added to the same note in full.
- 🔂 Same question again → one "wrong again" log line, and the review resets to tomorrow.
- 🚫 Turn off with Save mistakes as error notes in Settings.
🔁 Errors → Archive (mastery)
- 🆕 Quiz my errors writes a new version of each error. Same concept, new wording.
- 🚫 It never repeats a question: every version asked is logged in the note.
- 🎯 Types used: multiple choice, scenario, multi-select, true/false.
- 📅 Pick Due today or All active.
- 🔢 One new question per error. The count can't go above the errors found.
| Answer | What happens |
|---|---|
| ✅ Right, error was due | streak +1 → next check in 3 days, then 7 days (Settings: Days between checks) |
| ✅ Right, not due yet | ⚠️ Warning. Default: logged, not counted. Or Count anyway (+1), or red Skip → archive (asks twice) |
✅ 3rd right (streak: 3) |
status: mastered, moved to Errors/Archive/ (Settings: Correct answers to master, 1–10) |
| ❌ Wrong | streak: 0, status: new, again tomorrow |
- 📝 Misses are logged on the same error note. No new note.
- ⚙️ Setting Count a correct answer: only on due days (default) or any time.
Test my archive:
- 🆕 New version of each archived error.
- ✅ Right → 🗑️ Delete (to trash) or 📦 Keep.
- ❌ Wrong → back to active
Errors/,streak: 0, again tomorrow.
📊 Answer history + stats
- 🧾 Every answer (right or wrong) →
history.jsonin this folder. - 🔢 Saved: date, subject, note, topic, question, right/wrong. Max 5,000 (oldest dropped).
- 🚫 A new quiz on the same note skips its last 40 questions.
- 📊 Ctrl+P → Show my stats → score per subject, last 7 days, 5 weakest topics.
- 📄
<subject folder>/Stats.mdis rewritten when a quiz ends: score, last 7 days, errors due/active/archived, weakest topics, by quiz type. - ⚙️ Settings → Stats note = file name (empty = off). Folder quiz skips it.
- ⚠️ Stats come from the history. Keep answer history off → Stats.md shows ⏸️ paused and stops counting. Not a bug.
- ⚙️ Settings → Keep answer history (off = nothing saved) and Clear history.
- 💡 It is a history, not a question bank: questions are always new.
💸 Free limits (OpenRouter)
- ⏱️ 20 requests per minute.
- 📅 50 requests per day (1000/day after buying $10 of credits once).
- 🧮 One quiz = 1 request (note, folder, errors or archive).
- ✍️ Each short-answer check = +1 request.
- 🔁 Busy model (429/502/503) → it retries twice by itself (4s, then 8s).
- 🙈 A failed 429 may not show in OpenRouter Logs.
- 🔎 The error message shows OpenRouter's own reason.
🛠️ Errors and fixes
| Message says | Fix |
|---|---|
| Add your API key | Paste the key in Settings |
| Pick a model | Load free models → pick a :free one |
| 401 | Key is wrong or deleted → make a new key |
| 402 | Model isn't free → pick a :free model. Key's credit limit must be empty, not $0 |
| 404 | Model is gone → Load free models again |
| 429 | Free model busy upstream → pick another :free model. Or your limit → wait 1 min / tomorrow |
| Did not answer in 90 s | Free model stuck → try again, or pick another :free model |
| Response cut off | Ask for fewer questions |
| No usable questions | Try again, or pick another free model |
| Note too short | Note needs 80+ characters |
🔒 Privacy
Network requests (only 2, both to openrouter.ai):
- 🌐
POST https://openrouter.ai/api/v1/chat/completions: generates and grades questions. Sent only when you start a quiz (or press Check on a short answer / matching). Carries the note text, your settings for the quiz and your API key. - 🌐
GET https://openrouter.ai/api/v1/models: lists free models when you press Load free models in Settings. No note text. - 🚫 No telemetry, no analytics, no other servers.
Vault access:
- 📂 The file list (
getMarkdownFiles) is read only on your device: to find the notes of a folder quiz, to show the note picker in Quiz multiple notes, to find error notes, and for the folder / note search in Settings. File paths are never sent anywhere. - ✍️ Writes only inside your subject folders (error notes, archive moves,
Stats.md) andhistory.jsonin the plugin folder.
Data:
- 📤 The note text (max 24,000 characters) is sent to OpenRouter and the chosen model.
- 🚫 Don't quiz notes with private data.
- 🔑 The key is stored in
data.jsonin this folder, unencrypted. - 🙈
data.jsonandhistory.jsonare gitignored, so they never go to GitHub. - ☁️ iCloud still syncs it, so keep the key private.
📝 What's new
0.3.5
- 📚 New: Quiz multiple notes. Pick notes with search + checkboxes, or multi-select them in the file explorer and right-click.
- 🔒 README → Privacy now lists the 2 network calls and why the vault file list is read.
- 🛡️ Releases now carry GitHub artifact attestations (build provenance) for
main.js,manifest.json,styles.css. - 🤝 Added
CONTRIBUTING.md,CODE_OF_CONDUCT.md,SECURITY.md,ACCESSIBILITY.md, issue and PR templates.
0.3.4
- 🤖 Empty Question style → the AI infers the exam from the subject name + folder. Name + folder is enough to start.
0.3.3
- 🎨 Subject boxes get inner space, so their top edge is no longer cut off.
0.3.2
- 🎨 Settings → Subjects: the 3 boxes fill the row, and the Question style box is full width.
0.3.1
- 🔍 Settings → Subjects: the Folder and Rules note boxes search the vault as you type.
- 🆙 Needs Obsidian 1.4.10+. On older versions they stay plain text boxes.
0.3.0: generic and shareable
- 📚 No hardcoded exams. Subjects (name, folder, rules note, style) are set in Settings.
- 🏃 On first run, the CSM, PMP and CPMAI presets are added, but only if their folder exists.
- ⚙️ In Settings now: mastery count, days between checks, Errors/Archive folder names, folder-quiz skips.
- 🧾 Answer history in
history.json. A new quiz skips the note's last 40 questions. - 📊 Show my stats command.
- 📄
<subject>/Stats.mdis rewritten after each quiz. - ⏸️ History off → Stats.md says paused. Not a bug.
0.2.x
- 👀 Review missed replaces Retry missed. It shows the right answers and does not re-test.
- ⏱️ A request stops after 90 s, so it no longer hangs at "Generating".
- 🔘 Each question has its own option group, so a selection can't jump between questions.
- 📅 Quiz my errors shows due vs active counts.
- ⚠️ Not-due errors: Count anyway or red Skip → archive.
- 🔢 The question count is capped at the number of errors found.
- 📁 New: Quiz a folder, Quiz my errors, Test my archive, and the I was right button.
0.1
- 🧠 First version: note or selection quiz, 6 question types, hints, auto error notes.
🧑💻 For developers
- 📦 No build step:
main.js(CommonJS),manifest.json,styles.css. - 🌐 API:
https://openrouter.ai/api/v1/chat/completionsvia ObsidianrequestUrl. - 📐 Asks for JSON (
response_format: json_object). On 400 it retries without it. - 🧪 Pure helpers are exported on
module.exports._testfor unit tests. - 📚 Subjects live in
data.json(subjects: [{name, folder, rules, style}]). Presets are inPRESET_SUBJECTS. - ✅ Tested: helper unit tests (parse, normalize, grade, error notes, subjects, intervals, history) + a fake-vault flow.
- ⚠️ Not auto-tested: the UI and live OpenRouter calls.