The seven capstones (A to G) and the final capstone are specified in Lesson 24.1
(lessons/24.1-capstones.md, on the site at #/lesson/24.1): objective, what each builds on,
acceptance criteria, common failures and extensions, with the rubric and reporting standard they share.
They are specifications. The course provides no reference solutions for them; it provides the tools
(mjcourse.envs, mjcourse.control, mjcourse.experiment, mjcourse.stats) and the automatic part of
the acceptance check:
python -m mjcourse.capstone module:Class --kwargs '{"action_mode": "ee_delta"}' --object BODY
which runs Gymnasium’s checker and the course’s environment checks and prints the model fingerprint.
Put each capstone in its own folder here (for example a_drawer/) with its protocol, code, run records
and report.