docs(register): AI-process-language residual sweep round 3 — jargon leaked despite #352 + #362 audits #372

New issue

Closed

opened 2026-07-04 09:04:33 +02:00 by bosun · 0 comments

bosun commented

2026-07-04 09:04:33 +02:00

Owner

Finding

External cold-read round 3 flagged AI-process language ('Cold-Read', 'Surveyor', 'Bosun', 'ratify') still present in adopter-facing surface despite Herald's prior register-audit sweeps (#352 + #362).

Reviewer: 'It appears in docs/changelog enough that it makes the project feel internally process-driven rather than adopter-clean. Not fatal. But it does hurt trust.'

Recurrence framing

Same scoped-mechanism-vs-open-substrate class we've been catching all week. Two prior sweeps closed the primary vocabulary in workflow YAML + integration.md + ADRs, but residual leaks in:

CHANGELOG.md entries added since #352/#362 (any new entries carrying chamber-narration)
Newer docs (SECURITY.md, VERSIONING.md — created post #352)
Any docs sections not touched by the prior sweeps

Scope

Herald prose-craft surface — third register-audit delta:

Grep across all adopter-facing markdown for chamber-jargon (Cold-Read | Surveyor | Bosun | Pilot | Quartermaster | Herald | Engineer | ratify | ratified | ratify)
Per-instance judgment (keep-and-gloss vs scrub, same discipline as #352):
- Design-anchor 'Cold-Read' references (cold-read discipline is a real project artifact) → keep + gloss
- Chamber-name attributions ('Surveyor 41ce', 'per Bosun ratify') → scrub or replace with plain phrasing
- CHANGELOG entries: 'per Pilot Cold-Read' provenance → keep in commit messages (historical), scrub from user-facing CHANGELOG.md sections that ship in release notes
Fold #366 (docs/cold-read-prompt.md rename to cold-read-changelog.md + add cold-read-adopter.md) into the same PR since surfaces overlap
Cross-reference: this is #340 axis-3 or new #368 sweep — Herald judgment on which framing

Fix path

Bundled PR:

Grep-driven surface enumeration
Per-instance keep-and-gloss vs scrub decisions
Update CHANGELOG.md v0.2x.0 entries where chamber-narration reads as adopter-facing (not history-only)
SECURITY.md + VERSIONING.md any residual
Fold #366 rename work if timing aligns

Verification AC

Grep-zero on chamber-name attributions in adopter surface (README + docs/*.md except docs/internal/)
'ratified' + 'ratify' reads as maintainer-neutral (not chamber-attributed) or removed
Cold-Read references either kept + glossed OR scrubbed consistently
External cold-read round 4 finds no AI-process-language objection

Anchor

External cold-read round 3 (fresh ChatGPT session, 2026-07-04) verdict 'promising but not yet trustworthy'. Reviewer verified via execution (dry-run fresh-repo → v0.0.0 → feat → v0.1.0 all worked correctly). New findings surfaced despite round-2 delta closures. Operator ratified delta plan 2026-07-04. This tracker addresses one of the round-3 findings. Success criterion: subsequent external cold-read returns no further objections of this class.

BLOCKING v1.0.0 (should-fix upgraded to must — external explicitly named this class again after 2 prior sweeps).

## Finding External cold-read round 3 flagged AI-process language ('Cold-Read', 'Surveyor', 'Bosun', 'ratify') still present in adopter-facing surface despite Herald's prior register-audit sweeps (#352 + #362). Reviewer: 'It appears in docs/changelog enough that it makes the project feel internally process-driven rather than adopter-clean. Not fatal. But it does hurt trust.' ## Recurrence framing Same scoped-mechanism-vs-open-substrate class we've been catching all week. Two prior sweeps closed the primary vocabulary in workflow YAML + integration.md + ADRs, but residual leaks in: - CHANGELOG.md entries added since #352/#362 (any new entries carrying chamber-narration) - Newer docs (SECURITY.md, VERSIONING.md — created post #352) - Any docs sections not touched by the prior sweeps ## Scope Herald prose-craft surface — third register-audit delta: 1. Grep across all adopter-facing markdown for chamber-jargon (Cold-Read | Surveyor | Bosun | Pilot | Quartermaster | Herald | Engineer | ratify | ratified | ratify) 2. Per-instance judgment (keep-and-gloss vs scrub, same discipline as #352): - Design-anchor 'Cold-Read' references (cold-read discipline is a real project artifact) → keep + gloss - Chamber-name attributions ('Surveyor 41ce', 'per Bosun ratify') → scrub or replace with plain phrasing - CHANGELOG entries: 'per Pilot Cold-Read' provenance → keep in commit messages (historical), scrub from user-facing CHANGELOG.md sections that ship in release notes 3. Fold #366 (docs/cold-read-prompt.md rename to cold-read-changelog.md + add cold-read-adopter.md) into the same PR since surfaces overlap 4. Cross-reference: this is #340 axis-3 or new #368 sweep — Herald judgment on which framing ## Fix path Bundled PR: - Grep-driven surface enumeration - Per-instance keep-and-gloss vs scrub decisions - Update CHANGELOG.md v0.2x.0 entries where chamber-narration reads as adopter-facing (not history-only) - SECURITY.md + VERSIONING.md any residual - Fold #366 rename work if timing aligns ## Verification AC - Grep-zero on chamber-name attributions in adopter surface (README + docs/*.md except docs/internal/) - 'ratified' + 'ratify' reads as maintainer-neutral (not chamber-attributed) or removed - Cold-Read references either kept + glossed OR scrubbed consistently - External cold-read round 4 finds no AI-process-language objection ## Anchor External cold-read round 3 (fresh ChatGPT session, 2026-07-04) verdict 'promising but not yet trustworthy'. Reviewer verified via execution (dry-run fresh-repo → v0.0.0 → feat → v0.1.0 all worked correctly). New findings surfaced despite round-2 delta closures. Operator ratified delta plan 2026-07-04. This tracker addresses one of the round-3 findings. Success criterion: subsequent external cold-read returns no further objections of this class. BLOCKING v1.0.0 (should-fix upgraded to must — external explicitly named this class again after 2 prior sweeps).

bosun added the

priority/high

size/M

kind/docs

labels

2026-07-04 09:04:33 +02:00