Define the service boundary and ship target

This lab builds containerized answer API releasing image v2 at 5%→25%→100% weights. The human in the loop is release commander watching revision-sliced error rate. Scope is intentionally narrower than “make AI reliable”: you will prove one oracle — 5% canary healthy 20m → promote to 25%; guardrail breach → weight 0 in < 2m — and one invariant — promote only if canary error_rate ≤ baseline+0.5pp and groundedness ≥ baseline−1pp for two 10m windows. Record non-goals in your notes so a later change cannot silently expand authority. The incident mnemonic for the chapter is CANARY-PROMOTE-BLIND-6; design as if that ticket is already written and you are filling evidence.

Turn the oracle into a table: input fixture, expected observable, prohibited side effect, owner, latency/cost ceiling. Separate model taste from software correctness — transport, auth, parsing, and termination must be deterministic even when generated text varies. Primary metric family: canary_weight, error_rate_by_revision, grounded_rate_by_revision. Averages without a denominator or revision label do not gate release. Fake external dependencies in unit tests; live calls wait until fakes pass.

Before implementation, encode a failing check that would have caught automation promotes while scrape_up{revision=v2}==0. That failure is the pedagogical north star for later pages: contracts reject it, happy path never performs it, validation asserts it, failure-handling contains it, observability detects it, security-ops prevents privilege tricks around it, and mastery replays it in a drill. Endpoint under study: POST /answer.

Capacity note for planners: estimate peak demand on POST /answer and the cost ceiling for a failed retry storm. Write the abort conditions — unbounded spend, cross-tenant leakage, or inability to roll back — before you enjoy the first green test. Prefer synthetic fixtures shaped like production over anonymized production dumps you cannot share in class. When you are tempted to widen scope, re-read the oracle (5% canary healthy 20m → promote to 25%; guardrail breach → weight 0 in < 2m) and cut features that do not serve it. The teaching outcome is judgment under constraints: release commander watching revision-sliced error rate gets a trustworthy control, not a kitchen-sink framework. Keep the language of release decisions: promote, hold, or roll back — never “see if it gets better.”

Decide what will live in version control on day one: fixtures, contract markdown, and a failing test name. Write the cost ceiling as a hard number with currency and period. If the lab involves clusters, name the non-prod context you will use and forbid prod kubecontexts in scripts. Capture the baseline metric once before changing code so later gains are comparative. Refuse tools that hide the request path behind magic macros until the oracle is green on fakes. Your README section for this page should be five lines or fewer and still falsifiable. In this chapter the product is containerized answer API releasing image v2 at 5%→25%→100% weights, the human stakeholder is release commander watching revision-sliced error rate, and the incident id you design against is CANARY-PROMOTE-BLIND-6. Re-state the oracle in your notes — 5% canary healthy 20m → promote to 25%; guardrail breach → weight 0 in < 2m — and keep the invariant visible: promote only if canary error_rate ≤ baseline+0.5pp and groundedness ≥ baseline−1pp for two 10m windows. Track canary_weight, error_rate_by_revision, grounded_rate_by_revision as the scoreboard. Surface under change control: POST /answer. If you only have forty minutes, finish the fixture for automation promotes while scrape_up{revision=v2}==0 before polishing UI. Promotion language stays ternary: promote, hold, or roll back based on evidence, not hope.