Module 7 — The oracle problem
Status: Coming soon. Outline in _authoring/curriculum-outline.md.
This module is the conceptual hinge of the curriculum. The question every long-running agent has to answer: how does it know whether what it just did was good? Subjective oracles (asking another LLM) collapse under iteration. Objective ones (CI, lint, tests, seeded-issue detection) compound.
The lab builds ./oracle.sh <pr-branch> — a composite score from CI plus lint plus coverage plus seeded-issue detection. The artifact gets reused across every downstream module.
Strong candidate for publishing first, out of order. See the open structural questions in _authoring/curriculum-outline.md.