-
Notifications
You must be signed in to change notification settings - Fork 3
plat 208
PLAT-208 — report_human_inputs operator-identity gap is already closed by a shipped platform improvement, for every row going forward
| Coordination | Value |
|---|---|
| Assigned agent | unassigned |
| Ticket state |
resolved — informational closure record, no code change |
| Last synchronized | 2026-08-29 |
- Priority: P4 — no live defect requiring further work. The finding itself documents that the platform improvement it depends on has already shipped, and explicitly states there is no workflow-side action available.
-
Owner: N/A —
answered_by/answered_by_kind/answered_session_idare populated by the platform's own human-input-answering path, not by any workflow tool. -
Related:
db:report_human_inputs:operator-identity(confida-login, low), the finding this closes.PUL-07B2C9EC(referenced by the finding as a human-deference question that depends on this attribution) is a separate workflow-local concern, out of scope here.
"Materially changed by a platform improvement: recent decisions now carry
human_ui provenance and a real session id, so the original claim that no
approval in this table can be proven human is no longer true for anything
answered since 2026-08-12."
Seven rows answered 2026-08-12 or later carry
answered_by_kind='human_ui', answered_via='report_ui', and a concrete
answered_session_id. Only rows answered before that date remain
answered_by_kind='legacy_unattributed' with an empty session id — a
permanent historical gap for data that predates the provenance-tracking
improvement, not a live, ongoing one.
The finding's own next_check already states the correct action: "No
workflow change available — answered_by is populated by the platform,
not by this workflow. Treat pre-2026-08-12 rows as unattributable and stop
citing them as proof of human approval; reopen only if a NEW row appears
with answered_by_kind='legacy_unattributed'." That is a durable,
self-contained closure condition — this ticket exists to give it a PLAT
record and clear the harness row, not to add new work.
- No code change — the provenance-tracking improvement (
human_uiattribution with a real session id) is already live for every row answered since 2026-08-12; there is nothing further to build. - Did not retroactively attribute the pre-2026-08-12
legacy_unattributedrows — the finding is explicit that this is impossible by construction (the identity signal was never captured at answer time for those rows).
- Confirmed via the finding's own SQL reproduction against
db/db.sqlite'sreport_human_inputs: 7 rows since 2026-08-12 carryhuman_uiprovenance with a real session id; older rows arelegacy_unattributedby design, not by defect.
Auto-synced from docs/ on main. Edit there, not here.