Med math rationales were padded with unrelated drug facts (TSH and heart
rate for a levothyroxine tablet calculation) because the math prompt
asked for "one safety check" and the rationale format asked for at
least two bullets.
- Rationales take 1 to 4 bullets, each about what the question tests,
with fewer bullets preferred to padding.
- Med math rationales explain only the calculation: the conversion, the
method and an arithmetic check.
- The second review rewrites a rationale or "why" that brings in facts
the question does not test.
- The outside-chatbot instructions say the same.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Checked against a quiz generated from four neuro and mental health
lectures:
- The opener rewording now asks for the plain, active openings a real
exam uses and rules out inverted sentences ("Two days of vomiting have
left a client weak", "An order is received for…"). It steps in only
above 30% of a quiz, since the exam's most common opener was 24%.
- The finished set is tidied after review: plain stems, no closing
period on phrase choices, and first-person choices to a "Which
statement…" question in quotes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Modelled on a real 50-question course exam and the V1 question set:
- Stems are short and plain (usually 15-30 words, one paragraph, no
scene-setting), end in the exam's own question phrasing, and about a
third may ask directly about a fact the nurse uses.
- Bold and underline are stripped from generated stems (plainStem).
- Choices are concise and parallel: a term or value, a 4-12 word action
or finding, or quoted speech; explanations go in "why" (1-2 sentences).
- Distractors come from related facts in the material.
- Select-all items have 6 choices with 2-5 correct, mostly 3 or 4.
- The planner varies the kind of question instead of setting and role.
- Review, proofread, "improve questions" and the outside-chatbot
instructions follow the same rules.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Settings → Data → Import gets "Restore a full backup" in place of the V1
import row. The zip from "Download backup" is uploaded and replaces the
database, every setting and the uploaded files on the running server:
- the staged database is upgraded to this schema (newer backups are refused),
saved keys are re-encrypted for this server's key, interrupted jobs are
marked, and the restoring session is carried over;
- a pre-restore snapshot is written, running generations stop, and the file
is swapped under the open DB handle (DB.replaceWith); uploads and media are
copied into place and background work restarts.
The /api/import/v1 endpoint stays for the V1 cutover (docs/MIGRATION.md).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Stems no longer all open the same way: the writing rules, batch prompt and
planner ask for varied openings (age openings stay part of the mix), and when
one opener covers more than a quarter of a quiz the excess stems have only
their opening reworded, keeping every number, before the check.
- Rationales are asked for as Markdown (a bold key point, then short bullets);
older one-paragraph rationales are laid out the same way on display.
- The practice score in the quiz header is a plain "x/y correct", with no
percentage or colour.
- Fix a flaky assertion: the mock's answer-key fix is a no-op when the shuffle
already put the answer in place.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Settings -> AI: Planning, Writing, Python check and Second review each get
their own model, reasoning effort, temperature and output limit, shown under
Content generation. Unset choices follow Content generation, then the
default model.
- Shared quizzes say what Practice and Exam mean under the mode toggle.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Output limits leave room for reasoning (answer plus 8k-64k by effort, up to
the model maximum); a call cut off before answering is retried with double
the limit. At "max" effort, batches were being cut off at 16k tokens.
- Quizzes are planned first (one distinct angle per question), then written
in waves of four batches; med math runs alongside the first wave.
- Replacements for removed questions are planned and written in parallel.
- The Python check and the second review run at the same time each round.
Measured live (DeepSeek flash, max reasoning, 50 questions): 7 min 21 s with
no duplicates, against 20 min and 29 replaced duplicates without planning.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- The check can rewrite a whole question (garbled or contradictory stems,
missing data, two defensible answers); a key fix that changes nothing is
skipped.
- Removed questions are replaced with new ones, so a 50-question quiz stays 50.
- A second, independent model call reads the whole quiz for badly written
items, helped by a scan for leftovers such as "A 78 kg? No - an 80 kg".
- The quiz is checked again after any rewrite, replacement or open problem,
up to three rounds, before it is handed over.
- Reviewers see full rationales (a shortened one is marked as shortened).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Malformed model JSON is sent back for repair, then retried; a failed job
starts over once before it fails. Visitors can dismiss failed sets.
- Generated sets get a short distinct name and description; flashcards are
atomic and have no hints; standard study shows known/learning counts.
- Share links: "In progress" list, finished sets join the normal lists with a
"New" tag, plain progress steps, newest exam first.
- Share-link AI expires after a week, can be limited to the US or Arizona
(DB-IP Lite or Cloudflare headers) and is locked by abuse protection
(migration 8).
- The assistant never answers from general knowledge; MCP and web search
tools load without a manual test.
- Plain-text copy, no question count in exam mode, filler text removed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Sets visitors generate show in the owner's Drafts while being written, then
save into the class (and exam, for an exam link) automatically, as if
approved. If the save fails, the draft waits in Drafts for review.
- Deleting a share link keeps what was made on it; running jobs finish and save.
- Visitor flashcard sets go up to 75 cards (default 30); quizzes stay at 50.
- The visitor page links finished sets to the class items; the separate
AI-set view is gone.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Visitors of class and exam links can generate NCLEX-style quizzes (default
and maximum 50 questions, 25% select-all, Python-checked, optional 5 med
math questions) and flashcard sets from chosen class materials, once the
owner turns it on for the link.
- 4 quizzes and 2 flashcard sets per link per day; failed attempts do not
count; "Reset limits" in the share dialog gives a fresh allowance.
- Sets are drafts tied to the link (migration 7): listed for every visitor,
explainable, kept off the owner's drafts list, and reviewable or savable
from the share dialog.
- Generate gains "Include med math" with a count (default 5): exactly that
many dosage calculation questions at the chosen difficulty, spread through
the quiz, even when the material has none.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Interactive blocks are encouraged by default; Explain gets a reduced library
and never writes practice questions.
- Assistant tool steps breathe while running, show live stages from generation
and the Python check, and read in the past tense when done.
- The assistant asks before searching the web when class materials do not
cover a question; clarifying questions list every class's exams and never
ask about format (always NCLEX).
- Explain opens beside the question from 900px wide, moving the question left
while open and back on the next one.
- Generate dialog drops the practice set/exam switch (presets cover it) and the
Generate buttons lose their icon.
- Python check notes are green when every check passed, applied fixes neutral.
- Share pages group by exam, then Quizzes, Flashcards, Notes, Class Materials,
hiding empty groups; the class page uses the same order. "Documents" is now
"Class Materials".
- OpenRouter provider replaces the Ollama type; migration 6 turns Ollama
connections into OpenAI-compatible ones at /v1.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Built-in Parallel web search (keyless; optional sign-in for higher limits and research)
with its own settings page; OAuth sign-in (RFC 9728/8414/7591, PKCE, RFC 8707) for any
MCP server, with token refresh and pre-registered client support
- Friendly tool steps (icon, name, status; details on click) and ids stripped from results;
per-chat tool switches; taller thinking box
- Exams within classes: exam tabs, scoped lists, uploads, imports and Generate; exam-only
material for generation; class setting for what the groups are called
- Assistant can ask clarifying questions in an inline form before generating
- OpenUI blocks (charts, questions, flashcards, inputs) in assistant, Explain and share chats
- Explain beside the quiz on wide screens, grounded in the cited source pages
- Quiz results: percentage ring coloured by score band, rainbow at 100
- Share links: assistant button like the owner's, hidden during quizzes; documents readable
in the browser
- Live draft and library updates over server-sent events
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Generation
- Every quiz is NCLEX style; "Practice exam" writes 50 questions.
- Calculation questions only when the material involves calculations (max 5 per exam).
- Broad topics: the class's topics are listed in the prompt, a quick grouping call maps
narrow labels onto broad ones, and leftover variants are merged.
- Exact select-all count per batch; option order shuffled.
- Python check: at most 3 runs, fixed problems count as passed, cosmetic remarks ignored.
Assistant
- Text, tool steps and cards appear in the order they happen, with paragraph breaks
between steps.
- Optional thinking box: fixed height, follows the newest text, can be scrolled back.
- "+" opens a material picker listing recent items; items can be dragged in from class
pages; shorter placeholder.
MCP servers
- Streamable HTTP client; encrypted headers.
- Per-server instructions and switches for chat, Explain, generation and the quiz check.
- Settings UI.
Share links
- Optional "Explain this" and "Ask about this material" per link, off by default.
- Items are loaded on the server and checked against the share; no tools; rate-limited.
- Migration 4 stores the switches.
Study and documents
- Standard flashcard mode (optionally repeat missed cards until known) next to spaced
repetition.
- Streak visuals restyled in the app palette.
- Side by side / Original document views, with sandboxed previews of Word and PowerPoint
files.
Other
- Built-in embedding model baked into the image; sharp stubbed if it cannot load; real
load error shown in Settings.
- Python runner blocks network and process modules in JavaScript too.
- "View questions and answers" removed from the quiz start screen.
- Docs updated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Pyodide (Python 3.14, WebAssembly) in a separate Node process under the permission
model: read-only access to Pyodide's files, no writes, processes or network, empty
environment, JS bridges removed, time limit with kill and restart.
- Assistant tool run_python, which can load a draft or quiz as Python data.
- After quiz generation the model checks question count, type mix, correct answers per
type, duplicates and recomputed calculations with Python, then submits fixes that are
applied only if the question still validates. Switch in the Generate dialog.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Providers and pricing
- Add a DeepSeek provider: thinking mode with low/high/max effort, reasoning_content
replayed between tool steps, cache-hit usage, /models discovery and catalog prices.
- Fix "does not support tool calling": a rejected forced tool_choice (DeepSeek thinking
mode) now asks for the call in the prompt; only a genuine no-tools 400 is "unsupported".
- Seven reasoning-effort levels (Off to Max) mapped per adapter.
- Cached-input price per model, cache-write and DeepSeek off-peak aware cost estimates.
- Built-in bge-small embedding model (transformers.js) as the semantic search default.
Quizzes
- Numeric (type-the-answer) questions: units ignored, optional ± tolerance; editor,
runner, importer, templates and generation prompts.
- Quiz start screen instead of showing answers; navigator ring no longer clipped and
key legend removed; clock only for timed exams; 7-point letter grades.
Assistant and generation
- Assistant turns run on the server independently of the browser, with re-attach and
an explicit stop; drafts tray shows generations in progress.
- Per-chat model switcher, @-tagging of class materials, full-width assistant page and
a smaller assistant button.
- Editable generation instructions and built-in presets, separate from the chatbot
import instructions.
Other
- Dialog footer padding, keyboard hints removed, copy buttons work over plain HTTP,
class and folder shares include documents.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Drag-to-reorder for class materials, folders, quiz questions and deck cards. Pointer-based
so touch works, plus keyboard (arrows/Home/End) with screen-reader announcements.
- Timed exams get a wall-clock deadline. The server refuses late answers, submits overdue
exams with what was saved, and the countdown keeps running while the page is closed.
- Shared quiz sessions keep a frozen copy of their questions and flag later owner edits.
- Scanned PDFs are read in cached 8-page batches with real page numbers (pdf-lib), removing
the 30 MB whole-file limit. Compatible providers now receive PDFs as file parts.
- Semantic search: embeddings for OpenAI, Gemini, Ollama and compatible servers, background
indexing into chunk_vectors (migration 2), hybrid keyword + vector ranking for search, the
assistant and focused generation, and a Settings -> AI section.
181 tests pass locally and in the Docker test stage on Node 24.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
V1 remains intact on main and in history. This branch replaces it with a new
application: node:sqlite storage with migrations, scrypt auth and sessions,
content/SRS/attempt/stats services, document extraction and chunking, a
provider-neutral AI layer (OpenAI, Anthropic, Gemini, OpenAI-compatible,
Ollama), the generation pipeline and the tool-using assistant.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>