H-1

The verdict-of-record discipline puts the canonical TEST result in a Mattermost post under the veritas handle. H-1 is the rule that enforces who can post that record.

Without H-1, any agent could post a TEST PASS claim and the verdict register would be poisoned. The discipline closes that hole at the handle layer: only one specific Mattermost handle — veritas — may post TEST PASS or TEST FAIL. Posts from any other handle, even when the content reads like a verdict, are automatically invalid. The claim does not become verified by being declared verified by the wrong author.

The corollary is that submitting agents must not self-certify. Any agent running a local pre-flight check before sending a claim to Veritas must label its own report as [pre-flight]. This prevents the case where a substrate-lane agent runs axon-verify validate locally, gets a clean schema-validation result, and accidentally reports it in language that reads like a Veritas verdict.

H-1 includes a cat-1-reset semantics: when a category-1 field on a claim (the claim text, the asserted value, the cited source, the methodology) is edited after a verdict has been posted, the verdict is automatically invalidated. The claim must be re-tested. The earlier verdict stays in the channel as historical record, but the canonical-current is now the post-edit re-test result. Substrate-lane edits to category-2 fields (notes, internal tags) do not reset the verdict; only cat-1 edits do.

H-1 is what makes the verdict register usable as a register. The audit trail of who-said-what stays clean because the channel itself enforces the discipline.