- Practice tests on request in three modes (balanced, focused on the hard skills, manual pick) and simulations on request
(all the material, timed, no choice, repeatable). No plan, no window, no one-time rule. Old exercise tests 1-2 count as balanced, 3 as focused.
- Readiness moves only with handed-in tests, only up: number += GAIN x (grade - number), GAIN 0.6 simulation, 0.3 practice,
0.15 manual. Mastery and deepening no longer count. Levels, points to the next level, what a perfect test adds.
- Home: one compact block (number, level, days left; practice and simulation buttons with +N), still one screen.
The test page: how it goes up, simulation, practice modes, the trail of gains. A result opens with its gain and a level up.
- Teacher tabs show readiness, test counts and bests. countdown.py removed. OpenSpec change tests-on-demand; PRD updated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
AI review per question; the mistakes round shows where each question came from; a 15-question practice test on the weak topics
- Board review: one Gemini request per question (paced for the free tier), each result saved and shown as it comes, then a short
text-only summary. A stopping failure (quota, key) keeps what was checked; the teacher can check or re-check one question alone.
- Mistakes round: each question shows its number in the test and what was chosen there.
- Exercise test 3 (מבחן תרגול: נושאים לחיזוק): 15 questions from the skills that lost points in the test's exercise tests or simulation,
or are still being learned. Not part of the readiness score. On the test card and on a result.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
After a quiz, exercise test or the simulation: answer the wrong questions again
A mistakes round (kind mistakes_round) over the questions the handed-in attempt got wrong, in the lesson round's sheet: instant feedback,
never the right answer, counted as practice, the grade unchanged. Needs migration 20261122090000_m2_mistakes_round.sql.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Simulation: a bypass to start it on all the test's material when nothing is practiced yet
The tests expect board work's "work" fields and board functions; tests about other things count the board as written on. The home spec's "Nothing critical" scenario is back.
2 days ago
Refactor home screen layout to a single screen view with lesson tiles, nearest test strip, and new assignments button
- Updated home.js to change the layout to a grid of lesson tiles, each displaying stars and status indicators.
- Replaced the previous map structure with a compact design that fits within a single screen.
- Modified the nearest test display to a strip format showing days left and readiness.
- Consolidated new assignments into a single button that links to the submissions tab.
- Adjusted related tests to reflect the new layout and functionality.
feat: implement lesson rounds with instant feedback and star system
- Added a new lesson round feature that presents 10 questions from a lesson's skills, providing instant feedback for each answer.
- Implemented a star rating system based on performance in the lesson round, with conditions for earning 0 to 3 stars.
- Updated the home interface to display a map of lessons, showing progress and allowing easy access to the next lesson.
- Created database migration to accommodate the new lesson round functionality.
- Developed comprehensive tests for the lesson rounds and their integration with the home interface.
4 days ago
Only the simulation is timed: lesson quizzes and exercise tests have no time limit (#3)
* Path: heading reads מה נתרגל היום? instead of במה נתרגל היום?
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011XVGkAsva3pSS6u4MFvghb
* HF sync: check out without LFS
The image files' LFS contents were never uploaded to GitHub, so checkout
with lfs: true failed and the Space was never updated from GitHub. Check
out the LFS pointers and push them; the Space keeps the image contents it
already has.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011XVGkAsva3pSS6u4MFvghb
* Only the simulation is timed: lesson quizzes and exercise tests have no time limit
A lesson quiz and an exercise test now stay open until the student hands
them in and show no timer; the simulation keeps its countdown and is still
handed in when time ends. Attempts still store due_at (the database needs
one) but only a simulation is held to it. Updates the specs, PRD (QZ-2,
QZ-5, EXT-4, EXT-7, GRD-2) an…
6 days ago
Readiness parts as plain Hebrew rows with a short explanation each; archive readiness-parts-explained
- Introduced detailed specifications for exercise tests to enhance student practice under test conditions.
- Established a unified grading system across quizzes and tests, ensuring consistent evaluation.
- Defined the learning path structure, emphasizing skill mastery and navigation.
- Created lesson quizzes that allow retakes and provide feedback on skill performance.
- Implemented a readiness meter to assess student preparedness for upcoming tests.
- Added functionality for teachers to set and manage real tests within the app.
- Developed a rewards system to motivate student engagement and practice.
- Specified requirements for simulation tests to provide students with a rehearsal for real tests.
- Outlined skill mastery criteria to ensure accurate tracking of student progress.
- Designed a smart board feature for interactive problem-solving and submission of solutions.
- Enhanced the student home screen to streamline access to tasks and upcoming tests.
- Implemented a countdown plan…
