Dock

Most self-study drifts. Dock is the discipline that holds an AI partner and a curriculum accountable to each other: one word opens a session, another closes it, and between them lives an archive the assistant is not allowed to skip. The rule was born from a failure — two early sessions went undocumented and proved unrecoverable — and hardened into a promise: every session writes its own record, or the session did not happen.

The pedagogy is inverted

The machine designs the scaffold and reviews the output; the student writes the code, runs it, and reports back. A security frame instruments every fundamental — a set difference becomes a port-scan delta, a string cleanup closes a bypass before the comparison runs. Lessons name themselves as they surface and are filed as rules, each with a why and a where. When the protocol once referenced files that never existed, the drift was flagged rather than papered over — the rule revised to match reality, not aspiration.

What changed

Ten sessions logged without a gap since the discipline was set. The memory grew to five auditable, linked files; the knowledge base to twenty-three named principles. The AI stopped being a session and started being a system.