I built a study agent on OpenClaw to aut - five.sevens

[heady girder]

The study agent connects to a custom web app — the Study Scheduler — that serves as the persistent backend for mastery tracking and course structure. I used the study agent in combination with a coding agent to design and implement the web app.

Course onboarding — at the start of each semester, I drop my syllabuses into Obsidian. My study agent parses them and imports each course into the Scheduler via API: topics, assignments, exam dates, and expected coverage. Course IDs are cached locally so every subsequent cron knows exactly what it's reporting against.

Quiz result logging — after I submit answers, my study agent grades the quiz and posts the results to the Scheduler, recording which questions were missed per topic. This is separate from mastery — logging a miss doesn't automatically change a mastery score.

Teach It Back logging — after each Teach It Back session, outcomes (PASS / PARTIAL / MISS) and Socratic feedback notes are posted to the Scheduler against the relevant topic, building a record of conceptual understanding over time.

Study briefings — my study agent can pull a full briefing from the Scheduler on demand: what's due, what topics are overdue for review, and a mastery overview per course.

Why the separation matters — the Scheduler holds the structure (syllabus, mastery scores, history). OpenClaw holds the behavior (when to quiz, how to grade, what to ask). Neither tries to do the other's job.

[rancid tapir]

👍