College Cabinet

Problem

College tasks usually scatter across chats, spreadsheets, and notes: course weeks, lab rotations, deadlines, transfers, refunds, and reminders are easy to miss. College Cabinet keeps them in one local persistent engine, separating event from todo and routing courses, ledger, notes, habits, and briefings through separate specs instead of relying on chat memory or ad-hoc JSON.

How it works

• Safe startup: the first step reads SKILL.md; data must come from an explicit absolute path or an existing .college-cabinet/cabinet.db. If missing, it stops and asks for initialization or import.
• Engine-driven reads and writes: use query, snapshot, and doctor; writes take a revision, then apply a stable operation_id via apply. Conflicts require re-reading rather than overwrite retries.
• Verifiable delivery: reminders and briefings use claim, ack, and nack, and a write must be confirmed by re-reading the new ID, fields, and revision.
• Append-only history: ledger corrections and refunds are added as new records; deletions are soft, preserving audit trails.

Boundaries

It fits structured personal affairs like courses, schedules, budgets, notes, and reminders. It is not a general file search tool, database admin, or sandbox with no durable path. If the platform lacks a persistent scheduler, reminder config can be saved but external delivery cannot be guaranteed.