Bifrost Sprint 1 (2026-Q3) — Jira reconciliation, pass 9 (2026-07-19, F-21 correction)

Jira reconciliation — recheck (2026-07-19, pass 9)

Scope of this pass: a single correction to F-21, not a full re-pull. No new live-Jira fetch or sprint:reconcile run — the sprint 21367 digest is unchanged from pass 8 (142 issues). This pass exists because pass 8's code-evidence-gap read for Downgrade Webhook (BIF-8643) was incomplete, caught by a direct question in the same-day working session: "where do you check the code? It's already on hub-core."

Root cause. The code-evidence engine (and pass 8's manual follow-up) scopes its search per-initiative from the RFC body's known-slug mentions (sprint-resync §2d, resolution order 2). downgrade-webhook/rfcs/downgrade-webhook.md only names qontak-billing and moderator-be in its body — it never mentions hub-core — so every prior pass's code-evidence check for this initiative silently excluded hub-core from the search surface. This is exactly the "no code_repos declared → degrades gracefully, never guesses" behavior working as designed, but the initiative's own RFC pointed the scope at the wrong repo: the RFC's "qontak-billing" is a domain label (the billing feature surface), not the repo slug the code actually lives in.

Corrected evidence, hub-core, origin/production, fetched 2026-07-19:

• BIF-8811 (T1.1, triggers_downgrade column migration) — CORROBORATED: merged commit cc074768e8 / 76aec7c292, Merged in BIF-8811-bif-8643-t1-1-be-qontak-billing-downgrade-webhook- (pull request #8053), reachable from origin/production.
• BIF-8812 (T1.2, downgrade_events/downgrade_schedules migrations) — CORROBORATED: merged commit a467375fc4 / 8236e9e75f, Merged in BIF-8812-bif-8643-t1-2-be-qontak-billing-downgrade-webhook- (pull request #8054), reachable from origin/production.
• BIF-8813, BIF-8815, BIF-8698 — reconfirmed zero trace in hub-core too (in addition to qontak-billing/moderator-be already checked at pass 8) — consistent with their board status (To Do). Phase 2 (BIF-8813 schema follow-on, BIF-8815 logic/workers) is genuinely not started.

F-21 corrected verdict: Phase 1 (BIF-8811, BIF-8812) is CORROBORATED, not a code-evidence gap. The 2026-07-19 addendum's underlying claim — "Phase 1 schema is merged" — was substantively correct; its citation (PR #296, BIF-8813, qontak-billing) was a misattribution: wrong repo (should be hub-core) and wrong issue key (the real merged tickets are BIF-8811/BIF-8812, PR #8053/ #8054 — not BIF-8813, which is unmerged). Pass 8's "reopen, board/citation conflict, do not write up as resolved" call is accordingly superseded: Phase 1 progress can now be written up (once Addo/TPM confirms), while Phase 2 (BIF-8813/8815) stays correctly unwritten as not-yet-started. This is a narrower, more accurate finding than either pass 7's "0/3 code-corroborated, all in unchecked qontak-billing" or pass 8's "reopen, unresolved" — both were wrong about where to look, not about whether real work existed.

Durable fix applied: downgrade-webhook/README.md now carries an explicit code_repos: [hub-core, qontak-billing, moderator-be] override, so this initiative's scope no longer depends on the RFC body happening to mention the right slug. Future passes (including the real Sprint 1 close) will search hub-core automatically.

What this pass does NOT fix: it does not re-run the full digest/engine (unchanged since pass 8), does not touch any other finding, and does not write the corrected progress figure to downgrade-webhook/delivery/status/ yet — that write-up is proposed next, confirmation-gated, once Addo/TPM confirms per the ledger.

Downstream propagation worklist (this pass)

#	Finding	Canonical artifact to reconcile	Owning tool / skill
1	F-21 (corrected)	downgrade-webhook/delivery/status/<date>.md — Phase 1 (BIF-8811/8812) progress write-up, once Addo/TPM confirms the corrected attribution	skill step 6b, hand-applied
2	Durable fix	downgrade-webhook/README.md code_repos override — done, this pass	n/a