Should `--override` leave a trace of a forced illegal transition?

Gap: issue-0783d236d565 Blocking: no · Rank: 14 · Asked: 2026-07-29 · Answered: 2026-07-29 Authority: repo maintainer (chose from three options presented with trade-offs)

Question¶

Should --override leave a trace? Today a forced illegal status transition is indistinguishable from a legal one.

What the system does today¶

document_service.py:324-329 — status changes, nothing records that a Tier 0 rule was bypassed.

Proposed answer¶

Record it in frontmatter, or drop the flag and require an explicit two-step transition through a legal intermediate.

Why it matters¶

Coverage checklist: 'who can override a rule, and is the override audited?'

Answer¶

Yes to a trace, no to a record: warn loudly, store nothing. The escape hatch is deliberate and stays; what was missing was telling the operator, at the moment of the bypass, which rule they broke. An overridden: frontmatter field was rejected — no actors to attribute it to (adr-90e994d931cc), git already records the status change, and it would be permanent schema surface. Dropping the flag was rejected too: a document stranded by a schema change would have no path out but hand-editing, which the issue-6817ed1851e2 contract forbids. See issue-0783d236d565 resolution.

---

Migrated from the discovery question queue (Q-014); the queue itself now lives in this store.