Skip to content

Review Route Guide

Guide Maps

graph LR
  family["Reproducible Research"]
  program["Deep Dive DVC"]
  guide["Capstone docs"]
  section["Docs"]
  page["Review Route Guide"]
  proof["Proof route"]

  family --> program --> guide --> section --> page
  page -.checks against.-> proof
flowchart LR
  orient["Read the guide boundary"] --> inspect["Inspect the named files, targets, or artifacts"]
  inspect --> run["Run the confirm, demo, selftest, or proof command"]
  run --> compare["Compare output with the stated contract"]
  compare --> review["Return to the course claim with evidence"]

Use this guide when route choice itself is becoming noise. The goal is not to memorize targets. The goal is to match each state or release question to the smallest honest evidence surface.

Choose by question

If the question is about... Start here Escalate if needed
declared versus recorded stage truth make stage-summary make verify
promoted population facts make profile-summary make verify-report
promoted scoring behavior make model-summary make release-review
promoted inventory and release metadata make manifest-summary make release-review
borderline decisions near the threshold make threshold-review make release-review
concrete misclassifications make review-queue and publish/v1/predictions.csv make release-review
whether the promoted contract currently passes make verify make verify-report
saved contract evidence plus enforcement logic make verify-report make confirm
downstream trust in the promoted release boundary make release-review make confirm
remote-backed restoration after local loss make recovery-review make confirm
first-read orientation through the repository make walkthrough make tour
broader executed proof across repository state make tour make confirm

Escalation rules

  1. start with one summary command when the question is narrow
  2. use make verify when the question becomes contract validity
  3. use make verify-report, make release-review, or make recovery-review when the answer must survive later inspection
  4. use make confirm only when the whole repository story is under pressure

What this guide prevents

  • defaulting to confirm when a smaller route would answer the question more cleanly
  • using release review when the question is still only about stage truth
  • using one promoted metric as a substitute for profile, model, or record-level review
  • treating recovery review as if it replaced ordinary release review