Rendered from docs/acceptance-criteria/0002-every-required-section-of-every-governed-document-has-a-heading.md in the Headwater corpus. Every document on this half of the site is typed by the taxonomy the descriptor names: corpus.json.

Every required section of every governed document has a heading

Fit criterion

headwater check over this corpus reports no instance of section.required.missing.

The rule is an error rather than an advisory, so a strict run refuses a commit that would add one. .githooks/pre-commit runs the strict form, and the CI job runs it again on every pull request. So the criterion is re-established on every change rather than at one commit, which is the difference between this criterion and HW-AC-0001.

One thing the criterion does not state. The rule reads a heading, and it reads no word under one. A document with every required heading and nothing under any of them passes. That is the honest reading of what sections.require declares, and the remedy is a different rule rather than a stricter reading of this one.

Method

Test. engine/crates/check/src/sections.rs carries the rule and no case table of its own. The golden fixture in engine/crates/check/tests/fixtures.rs proves it, over the tree at engine/crates/check/fixtures/check/. check/spec/08-contract-met.md writes every heading its kind requires, and check/spec/09-contract-missing.md omits one. Both feed the recorded report at engine/crates/check/fixtures/check.report, and the test compares that report byte for byte.

Two commands produce what a reader inspects.

cargo test -p headwater-check --test fixtures --manifest-path engine/Cargo.toml --locked
headwater check --root . --strict

The first holds the rule to the two fixture documents and the report they produce. That is what makes the method test rather than inspection: the fixture documents decide the verdict, not a person who reads a tree. The second command runs the rule over this corpus and reports what it found.

The two are one method and not two. A rule that could report nothing would leave the second command silent over any corpus at all. A green run and a clean corpus then look the same. The fixture report is what separates them.