a record authored straight into resolved never invokes capture resolve so the grounds obligation reaches only records that pass through the verb

a record authored straight into resolved never invokes capture resolve so the grounds obligation reaches only records that pass through the verb

Found by the itd-179 fidelity audit and verified at integration HEAD.

itd-179 makes capture resolve|wontfix|promote refuse without grounds. That obligation reaches exactly the records that pass through those verbs.

Most do not. Measured at HEAD: 689 of 715 records in resolved/ carry no ## Grounds section, and all 689 carry a resolution: frontmatter key -- authored directly into the terminal folder rather than transitioned into it. The auditor's scoped measurement is sharper: 95 records entered resolved/ after the refusal landed on this branch, and 69 of those carry no grounds.

The distinction that matters: the gate is not bypassed, it is never invoked. A refusal on a verb cannot bind a workflow that does not call the verb, and authoring a record straight into resolved/ is a normal thing to do in this repository -- the branch that shipped the obligation is itself the largest producer of grounds-less terminal records.

Related but distinct from iss-2608301747006182, which observes that no GATE requires a terminal record to carry grounds. This record is about why arming that gate would be a bigger change than it looks: it would refuse the majority of the existing corpus, and the population it would refuse was created by the ordinary working practice of this repo.

RULED 2026-08-31: a forward-only gate. A terminal-folder record must carry a grounds entry if it was created after the gate is armed; the existing corpus is exempt.

The precedent is this cycle's own: the intent-side grounds gate was promoted forward-only and left 36 intents NOT READY without refusing any of them. The same shape applies here, and for the same reason — the population that would be refused was produced by the ordinary working practice of this repository, so refusing it retroactively punishes the practice rather than changing it.

The cutover is the ARMING COMMIT, not a date. A date written in prose is exactly the fact itd-195 says not to state: it cannot fail, and it decays the moment anyone misremembers when the gate landed. Whoever arms the gate records the commit it was armed at, and the gate compares against that.

Rejected, with reasons kept: narrowing the claim and adding no gate at all, which is honest but leaves roughly three-quarters of new terminal records escaping an obligation the intent's press release implies is universal; and forcing the verb to be the only door, which would make the obligation genuinely universal but needs 689 records migrated and fights how this repo is worked daily — too large a change for what this finding establishes.

Needs a record home before it is built. It is a capability change on itd-179's surface, so it is either that intent's follow-up or a new one; that routing is a decomposition hand-run nobody has done.

Grounds

• declined: the grounds gates are parked as advisory until the reading work is rethought, so no terminal-record grounds gate will be armed; this would be wrong if that parking were lifted and the forward-only gate re-ruled