Integration handoff

Version 2026-10-14.3 retains the technical-review fixes and adds guided capstone bridges, learner-code checks, inspection/debugging practice and anchored scoring. This package contains content and tested reference implementations only. No production route, enrolment state, learner-progress system or account behavior has been changed.

Map content to the existing site

1. Keep the course identity python-ai-data and existing public overview URL.
2. Use the ordered modules in course-manifest.json. Preserve prerequisite order and lab-profile labels.
3. Render module Markdown with syntax-highlighted, selectable code. Keep worked-script downloads beside the examples.
4. Keep exercises and solutions separate. A reveal control must not imply a solution was independently completed.
5. Publish dataset descriptions with the CSVs so users understand synthetic origin, row meaning and intentional defects.
6. Preserve SVG alt text and accessible title/description. Do not replace diagrams with unlabeled screenshots.
7. Keep the current public planned/enrolment copy until an authorized release changes it. Lab practice checks are not course completion evidence.
8. Treat the lab URL as an external practice destination. Do not invent profile-query URLs or integration APIs; the verified workflow is selecting the named profile in the UI.

Checks in the final environment

• Every module, source link, exercise, solution, dataset and download opens for a normal learner account or public visitor as intended.
• The full code is copyable without prompts, line numbers or invisible formatting characters.
• Keyboard navigation reaches headings, practice links, downloads and any answer-reveal controls.
• Diagrams are readable at mobile widths, with a zoom or text alternative if necessary.
• The browser Data profile runs the capstone data stage, and Charts runs the separate list-based chart stage.
• Browser download wording matches actual behavior. Generated runtime CSV files are not promised to persist or export automatically.
• Local requirements and test results are exposed as tested versions, not universal compatibility claims.
• A human beginner pilot checks pacing and usability before the course is described as learner-validated.

Known technical limitation to preserve in release notes

The tested lab Charts profile rendered downloadable charts but emitted Matplotlib deprecation warnings in the browser runtime. The course uses explicit PNG output to avoid the separate non-interactive plt.show() warning. The browser runtime warnings do not prevent the tested output, but should be reviewed as a lab maintenance issue before promising warning-free exercises. No lab code was changed in this task.