Changelog

Unreleased

Added

• Configurable completion modes: on submit, on view after a delay, and manual completion through window.markComplete().
• Optional AI content generation and refinement in Studio through an OpenAI-compatible endpoint or custom provider callable.

Security

• AI generation now checks Studio write permission and fails closed outside a supported Studio runtime.
• The Test Grading panel escapes feedback, constructs multi-field controls with DOM APIs, and accepts preview messages only from its own sandboxed iframe.
• Generation prompt and refinement content sizes are capped.

Fixed

• Staff detection now uses the Open edX edx-platform.user_is_staff attribute, restoring the staff-only debug panel.
• Generated activities enable grade publication.
• Refinement preserves the display name and grading configuration when the generated result does not replace them.
• AI response parsing handles code containing triple backticks more safely and validates multi-field grading metadata.
• The default non-streaming model timeout is 120 seconds, with clearer handling for non-JSON responses.
• The test suite uses supported XBlock request and fragment APIs.

1.0.0 — 2026-04-03

• Added HTML, CSS, and JavaScript activity rendering.
• Added the submitInteraction(data) browser API and per-learner interaction state.
• Added JavaScript-controlled grading, single-answer matching, and multi-field partial credit.
• Added Studio grading controls and Test Grading.
• Added completion tracking, configurable feedback, previous responses, a staff-only debug panel, and masquerade support.
• Added support for Django 4.2 and 5.2 and Python 3.11 or later.

0.1.0 — 2025-08-05

• Published the first release to PyPI.

For commit-level history, see the GitHub repository.