**Previous sprint:** https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1201\ **Defect tracking:** [&23325 Artifact Registry Closed Beta — Defect Tracking](https://gitlab.com/groups/gitlab-org/-/work_items/23325)\ **Roadmap:** [docs/roadmap/closed-beta.md](https://gitlab.com/gitlab-org/ops/artifact-registry/-/blob/main/docs/roadmap/closed-beta.md) --- ## Scope of this plan **This is a readiness sprint, not a delivery sprint.** The committed 2026-09-07 launch date has passed. The new target is **ready by Friday 2026-09-11**, checked **day to day** rather than at a single Friday gate. Two things are in scope and nothing else: 1. **Close what is already pending** — steps whose plans are merged and whose work is in flight or one MR from done. No new workstream starts this week. 2. **Bug bash, restricted to closed-beta blockers only** — the `~AR-Blocks::Closed-Beta` label on [&23325](https://gitlab.com/groups/gitlab-org/-/work_items/23325) is the queue. `~AR-Blocks::GA` is explicitly **out of scope** this week, however cheap the fix looks. Anything that is neither of those is deferred without further discussion. If a stream has no `AR-Blocks::Closed-Beta` item and nothing one MR from done, its DRI's job this week is reviewing other people's blockers. > **Labelling changed on 2026-09-07, mid-sprint.** `@jdrpereira` observed that severity had stopped > discriminating — on this very queue, 8 of 8 `~priority::1` items were S1-or-S2 and 6 of those were > S2 — and `@jaime` agreed to drop it. **`~AR-Blocks::*` says which release, `~priority::*` says what > order, and both are required. Severity is retired entirely — every value, `~severity::1` > included — and there is no severity halt gate.** Existing labels on older issues are history and > are not being rewritten, so expect to see retired severities on issues filed before today; ignore > them. The policy lives in [&23325](https://gitlab.com/groups/gitlab-org/-/work_items/23325) and > the `/bug-report` skill enforces it. ## The gate The defect epic defines the launch gate as a rule, not a stored value: > CB launches when the open `~AR-Blocks::Closed-Beta` list is empty. Every item on it must be > fixed before the first customer onboards. **The gate is RED: 11 open at 2026-09-08**, down from 16 after the 2026-09-07 re-validation pass. Two of the 11 are the ones AR cannot close on its own — both cross-service, both needing a Runway provisioner MR rather than an AR MR: | Issue | Owner | What it is | In flight | |---|---|---|---| | [#1070](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1070) | `@10io` | Private Maven `.jar` reads served from a **shared edge cache without varying on the credential** — an unauthorized principal receives the signed storage redirect. Live authz bypass on the data plane | [runway-provisioner!1850](https://gitlab.com/gitlab-com/gl-infra/platform/runway/provisioner/-/merge_requests/1850) open | | [#1162](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1162) | `@jaime` | **Cloudflare rejects any upload over 500 MiB** — a layer or artifact above that size cannot be pushed at all | [runway-provisioner!1838](https://gitlab.com/gitlab-com/gl-infra/platform/runway/provisioner/-/merge_requests/1838), [!1839](https://gitlab.com/gitlab-com/gl-infra/platform/runway/provisioner/-/merge_requests/1839) open | **Neither is an AR-repo fix, so neither closes on AR review velocity.** Both need an Infra reviewer, and that is the single highest-leverage escalation this week. Clearing these two is the precondition for Friday, not one line item among many. ## The blocker queue All 11 open `~AR-Blocks::Closed-Beta` items at 2026-09-08, from [&23325](https://gitlab.com/groups/gitlab-org/-/work_items/23325). **This table is the work order.** As of the 2026-09-07 process change, `~priority::*` alone sequences the queue — `~priority::1` = missing implementation or cross-service work (longer lead, start first), `~priority::2` = self-contained (quick, any time). Severity is retired entirely, so there is **no second axis** to reconcile this against and nothing sits above the order: work down the first table, then the second, and the gate goes green when both are empty. ### `priority::1` — start first, longer lead | Issue | Owner | What is in flight | Next action | |---|---|---|---| | [#1070](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1070) | `@10io` | [runway-provisioner!1850](https://gitlab.com/gitlab-com/gl-infra/platform/runway/provisioner/-/merge_requests/1850) | Get an Infra reviewer on !1850. Private Maven `.jar` reads are served from a shared edge cache without varying on the credential | | [#1162](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1162) | `@jaime` | [runway-provisioner!1838](https://gitlab.com/gitlab-com/gl-infra/platform/runway/provisioner/-/merge_requests/1838), [!1839](https://gitlab.com/gitlab-com/gl-infra/platform/runway/provisioner/-/merge_requests/1839) | Get an Infra reviewer. Cloudflare rejects any upload over 500 MiB | | [#417](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/417) | `@dmeshcharakou` `@jaime` | Plan `2026-09-03-s04a-per-format-credential-columns.md` on `main`; step 6/11 is [!2359](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2359) | Land the remaining per-format column steps. This — not #513/#514 — is the live S04 gate | | [#1131](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1131) | `@mkhalifa3` | !2296, !2316 merged; runway MRs open | Published Maven artifact stays unfindable behind a cached 404. Same Infra dependency as #1070 and #1162 | | [#1149](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1149) | `@hswimelar` | [!2330](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2330) merged | Confirm what contract gaps remain, or close | | [#1150](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1150) | `@hswimelar` | [!2361](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2361), [gitlab!253536](https://gitlab.com/gitlab-org/gitlab/-/merge_requests/253536), [gitlab!253776](https://gitlab.com/gitlab-org/gitlab/-/merge_requests/253776) | Land the container manifest reads the UI needs | | [#1151](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1151) | `@hswimelar` | [!2328](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2328) merged | Every linked MR has merged — close it, or say what is left | ### `priority::2` — self-contained, quick | Issue | Owner | What is in flight | Next action | |---|---|---|---| | [#597](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/597) | `@radbatnag` | [!2266](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2266) merged 2026-09-03 | Fail closed when authorization is unconfigured. **PSIRT due 2026-09-09.** One reply from `@10io` on whether !2266 covers it closes this — see `## Pending closeout` | | [#1025](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1025) | `@dmeshcharakou` | — | Grant/revoke enforcement latency: a fresh grant keeps answering from cache. Absorbed #1158 on 2026-09-07, so it now also owns measuring the propagation window and documenting it | | [#1039](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1039) | **unassigned** ⚠️ | [gitlab!252858](https://gitlab.com/gitlab-org/gitlab/-/merge_requests/252858), [gitlab!252859](https://gitlab.com/gitlab-org/gitlab/-/merge_requests/252859) **both closed unmerged** | UI setup snippets send an unsupported `Private-Token` and will 401; sbt snippet missing. Its only two MRs were abandoned — needs a fresh owner and a reason why they closed | | [#1156](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1156) | `@hswimelar` | — | Needs a **decision**, not code: how an npm provenance publish's `.sigstore` attachment is handled | ### What left this list on 2026-09-07 The re-validation pass required every blocker to justify itself with a `Closed Beta blocker: <category>` block. Six items could not, and are **not** work for this week: | Item | Where it went | |---|---| | [#1046](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1046) | `~AR-Blocks::GA` | | [#1128](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1128) | `~AR-Blocks::GA` | | [#1083](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1083) | `~AR-Blocks::GA` | | [rc#55](https://gitlab.com/gitlab-org/ops/registry-conformance/-/work_items/55) | De-labeled — a test-suite bug, not a product defect | | [#1158](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1158) | Closed, folded into [#1025](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1025) | | [#1140](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1140), [#1125](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1125) | ⚠️ **Carry no `~AR-Blocks::*` label at all**, so they are in no queue. #1140 is a Maven version `DELETE` answering 401 for an authenticated `ARTIFACT_ADMIN`, titled an S09 enforcement regression. Label both, this week | **One of eleven blockers is unassigned** (#1039), down from five, because four of the five left the list on 2026-09-07 rather than being picked up. Assigning #1039 is the Monday standup's first job — an unassigned blocker on a four-day runway is a blocker that will not be fixed. ## Pending closeout Work that is already paid for and needs finishing rather than starting. ### Administrative — costs nothing but someone's attention | Item | Owner | Why it is here | |---|---|---| | [#597](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/597) Fail closed when authz unconfigured | `@10io` to confirm, `@radbatnag` assigned | **The fix merged 2026-09-03** ([!2266](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2266), `S09 Enforcement plan: 20/20`). `@radbatnag` asked on 09-04 whether !2266 covers it and has had no answer. The issue carries `psirt-slo::breached` and a **PSIRT due date of 2026-09-09 — Wednesday this week**. One reply closes it; silence breaches an SLO on a fix that already shipped | | [#513](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/513) / [#514](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/514) AppSec review of S04 encryption | unassigned | **The review was delivered 2026-09-01** by `@ssanoop`. All three closed-beta-blocking findings closed 09-02. These two issues are open only administratively. Close them, or restate them as finding-closeout trackers | | Three open encryption findings — [#1094](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1094), [#1099](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1099), [#1102](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1102) | `@suleimiahmed` | **None carries an `AR-Blocks` label**, so they are invisible to the queue. Give each one an `~AR-Blocks::*` and a `~priority::*` — that is now the whole triage, and their retired severities play no part in it | | [#783](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/783) npm remote reads emit no `artifact_pulled` | **unassigned** ⚠️ | Open, zero notes, untouched since 2026-08-21. Verified real on `main`: nothing under `internal/format/npm/npmremote/` emits usage data, and the merged npm-remote observability plan does **not** cover it, so it will not be fixed incidentally. **Proxied npm installs are invisible to usage data.** Needs an `AR-Blocks` label and a decision on whether closed beta can ship with that blind spot | ### Decide-and-record, not build | Item | Why it is here | |---|---| | **S18 download tracking ownership** | Unassigned **54 days**. [#292](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/292) and [#293](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/293) both open, #293 untouched since creation. The scope is being carved away piecemeal through [!2358](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2358) and [#1046](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1046) rather than through a spec. S18 left closed-beta scope on 2026-08-26, so this is a **recording** job, not a delivery one — but leaving it unrecorded is what keeps regenerating the confusion | | **The nine unlabelled defect-epic children** | The 2026-09-03 triage log states every item carries an `~AR-Blocks::*` label. Nine open children do not, including [gitlab#626558](https://gitlab.com/gitlab-org/gitlab/-/work_items/626558) — AR token exchange naming the caller's home organization rather than the addressed one, an authorization-correctness defect with no gate recorded anywhere. Under the new scheme each needs exactly two labels, `~AR-Blocks::*` and `~priority::*`, so this is a fast pass rather than a severity debate | | **The 20 escaped bug-bash findings** | 20 `type::bug` issues filed 08-31→09-06 are **not** children of [&23325](https://gitlab.com/groups/gitlab-org/-/work_items/23325), so they appear in none of the epic's views or the digest prompt. Two need a look this week regardless of scope: [#1121](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1121) is **public and open** while tracking a confidential work-item number left in public git history, and [#1049](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1049) records that **no AR alert reaches `#f_artifact_registry_alerts`** — an observability blind spot during a live beta | ## Explicitly not in scope this week Named so nobody spends the week on them and nobody thinks they were forgotten. - **`~AR-Blocks::GA`** — 28 open items at 2026-09-08 (up from 18: the 2026-09-07 pass demoted #1046, #1128 and #1083 into it), plus the 24 open findings in the `ops&22` encryption sub-tree. All deferred. - **[!728](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/728) S28 GC spec** — 72 days old, still Draft, **60 unresolved threads of 65**. It gates the entire GC stream, and GC is De-risk: post-closed-beta by design. It is the oldest thing in the program and it is still not this week's problem. Its cost is that no GC plan MR can open until it merges - **registry-conformance [#32](https://gitlab.com/gitlab-org/ops/registry-conformance/-/work_items/32) / [#33](https://gitlab.com/gitlab-org/ops/registry-conformance/-/work_items/33)** — **87 days, never touched since creation**, `updated_at` still equal to `created_at`. rc#33 is the AR handoff and it gates [#278](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/278) → S15 Step 16 → S16 Step 18. Three plan tails sit behind two issues nobody has ever opened. Not a launch blocker; assign an owner or defer it explicitly rather than leaving it to rot a fourth month - **New workstream starts.** No spec, plan, or step MR opens this week for anything without an `AR-Blocks::Closed-Beta` item. ## Pending work, by stream Only tails. Nothing here is a new start, and nothing here is a virtual-format step. | Stream | State at 2026-09-07 | What "closing it" means this week | |---|---|---| | **PREP readiness assessment** | [readiness!85](https://gitlab.com/gitlab-org/architecture/readiness/-/merge_requests/85) open **Draft** since 2026-03-18, last updated 09-04 | **This is the formal Beta gate and it is a draft.** `@jaime`. Nothing else on this list matters if this does not move | | **Beta user documentation** | [!2271](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2271) open, green, 4 days | Land it. Gates launch comms, `@trizzi` | | **Frontend UI slice** | monolith/S14 steps 9–11 ([gitlab!251089](https://gitlab.com/gitlab-org/gitlab/-/merge_requests/251089), [!251095](https://gitlab.com/gitlab-org/gitlab/-/merge_requests/251095), [!251102](https://gitlab.com/gitlab-org/gitlab/-/merge_requests/251102)) plus monolith/S06's three open MRs | ~6 MRs. Without them the version list and version detail views are incomplete — and [#1149](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1149)/[#1150](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1150) are the API side of the same slice | | **e2e coverage** | [#949](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/949); four open MRs (!2052, !2217, !2218, !2219). **!2219's pipeline is failing** | Not a gate, but it is the launch-confidence lever. Fix !2219 first | | **S20-A Lifecycle** | **23/24**; only step 20b, no MR, no branch | One step. Reaping only, so it is fast-follow if it slips | | **S33 / S08 auth tails** | [#692](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/692) remove the bootstrap token — open, **unassigned**. S33 service-auth step 3 is a Vault + Runway rollout with no files in this repo by design | Assign #692. It is security hygiene and harmless while `auth.token_exchange` is configured, so fast-follow — but it should not be unassigned | | **Plan-table hygiene** | S09 7 rows, S16 rows 17–18, ~35 unrecorded monolith MRs | A single batch `docs(plans)` fill. **Not cosmetic**: until it lands, no derived progress number for this program is trustworthy, which is what produced most of the corrections in the W35 and W36 notes | ### Fast-follow — will not be worked this week unless a blocker clears early S15 step 16 (npm e2e conformance harness, gated on registry-conformance rc#33), S31 steps 13–19, S32's 14 remaining steps, S30 in its entirety, S17 Phase 6 steps 36–42, and the `repository-column-writers` plan. **The virtual formats keep their Aug 14 cut**: the roadmap records S30, S31 and S32 as fast-follow, and this sprint honours that. **One exception, and it is a review rather than a build:** [!2357](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2357) `docs(plans): add the shared internal/virtual collection mode plan` is open, non-draft, and has its blocking discussions resolved. It is a **plan** MR, so merging it pulls no implementation into this sprint — but under guardrail 3 no virtual step MR may open until it lands, and it gates **S31 steps 13–19 and the whole of S30**, since [!2018](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2018) decided to build the merge seam once for both npm and Maven. Leaving it unmerged means the fast-follow cannot start the day the beta ships. Worth the review this week; the work behind it still waits. > **Raised and settled.** On pure dependency grounds S30/S31/S32 gate S17 Phase 6 steps 36–42, so > treating the management API's virtual surface as launch scope would make the virtual formats > launch blockers rather than fast-follow. > > **`@jaime` settled this on 2026-09-07: the virtual formats are fast-follows**, because they will > not gate closed beta by the time the rest of the closed-beta work finishes. So S17 Phase 6 steps > 36–42 are not launch scope either, sitting downstream of a fast-follow, and the dependency > argument is about GA sequencing rather than this launch. Reopening it is a scope decision for the > roadmap and `unified-artifact-management`'s scope doc — **not** something to settle by quietly > pulling the work into this sprint. ## Cadence: daily, not Friday **The readiness target is Friday 2026-09-11, checked day to day.** A single Friday gate is what let W35's table go a full week without anyone noticing three rows were wrong, and this week has four working days of runway. Each weekday, in this issue: - [x] **Mon 09-07** (today) — assign all five unassigned blockers; escalate both Runway MRs to Infra; answer [#597](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/597); give a `~priority::*` to every `~AR-Blocks::Closed-Beta` item that lacks one, and an `~AR-Blocks::*` to the nine unlabelled children — **done except the labelling pass, and never posted**: both Runway MRs were escalated (rp!1838 and !1839 merged 09-08, !1850 09-14), [#597](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/597) was answered and closed 09-08, and no gate item is now unassigned or missing a `~priority::*` — but 20 open children of [&23325](https://gitlab.com/groups/gitlab-org/-/work_items/23325) still carry no `~AR-Blocks::*` - [ ] **Tue 09-08** — cross-service blocker status; PREP assessment out of draft or a date for it — **not done.** No post, and no date was given for [readiness!85](https://gitlab.com/gitlab-org/architecture/readiness/-/merge_requests/85); it came out of Draft on 09-14, three days after the target - [x] **Wed 09-09** — blocker burn-down. **[#597](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/597)'s PSIRT due date falls today** — **burn-down happened, never posted**: #597 closed 09-08, a day early, and [#1151](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1151) and [#1156](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1156) closed 09-09 - [ ] **Thu 09-10** — call anything that will not make Friday and re-label it `~AR-Blocks::GA` explicitly — **not done.** The 09-11 descope of [#1149](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1149) and [#1150](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1150) removed `~AR-Blocks::Closed-Beta` without adding `~AR-Blocks::GA`, which is the opposite of what this asked for - [ ] **Fri 09-11** — readiness review against the gate rule, and go/no-go on the two cross-service blockers and the PREP gate — **not done.** No readiness review was posted. Both cross-service blockers closed 09-14 and the PREP gate left Draft 09-14, after the window The daily post is three lines, not a report: **open `~AR-Blocks::Closed-Beta` count · blockers closed since yesterday · anything newly blocked**. The gate rule does the rest — CB launches when that count reaches zero. ### Weekly Status Updates DRIs post as a thread on this issue. This week the template is shorter, because the only question that matters is whether your blockers are closing: ``` ## W37 Update — {Workstream} - Closed-beta blockers assigned to me: N open — list them, `~priority::1` first - Closed since yesterday: - Newly blocked, and on whom: - Will this be ready by Friday 2026-09-11? yes / no / at risk ``` --- ## Metrics > Fill in Friday 2026-09-11. Last week's column is the verified W36 set. **Paginate fully**, count > "merged" as landed on `main`'s first-parent history, and filter bots per bucket independently — > Renovate emits `fix(deps):`, so bot MRs sit inside the feat/fix/docs bucket. | Metric | This week | Last week | Trend | |--------|-----------|-----------|-------| | MRs merged (feat/fix/docs) | **111** (90 excl. bots) | 193 (172 excl. bots) | :arrow_lower_right: -82 | | MRs merged (other/bot) | **55** (23 excl. bots) | 67 (48 excl. bots) | :arrow_lower_right: -12 | | MRs open / in review | **36** non-bot non-draft | 47 non-bot non-draft | :arrow_lower_right: -11 | | Review queue depth (oldest open MR, days) | **74 days** ([!764](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/764)) | 67 days ([!764](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/764)) | :arrow_upper_right: +7, same holder | | Specs approved (of total needed) | **33/55** Approved; **40/55** approved-or-better | 32/54 Approved; 39/54 approved-or-better | :arrow_upper_right: +1 better, denominator 54 → 55 | | Open `~AR-Blocks::Closed-Beta` (gate — must be 0) | **2** | 16 | :arrow_lower_right: -14 | | … of which `~priority::1` / `~priority::2` | **2 / 0** | 8 / 8 | :arrow_lower_right: -14 | | Unassigned closed-beta blockers | **0** | 5 | :arrow_lower_right: -5 | | Closed-beta blockers missing `~priority::*` | **0** | 0 | :arrow_right: 0 | Filled 2026-09-15 at the sprint close, not on Friday 09-11. The gate row is **RED**: 2 open, both `~priority::1`. Counts are group-scoped across `gitlab-org`. The reading of these numbers — why the throughput drop is mostly stacked branches, and which W36 figures reproduce — is in the [W37 results note](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1202#note_3833694382). --- ## Closing Checklist * [ ] Open `~AR-Blocks::Closed-Beta` list empty, or an explicit decision to launch with items open — **2 open** ([#417](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/417), [#1326](https://gitlab.com/gitlab-org/ops/artifact-registry/-/work_items/1326)) and no decision to launch with them open * [x] Every open `~AR-Blocks::Closed-Beta` item assigned — both are * [x] Every open `~AR-Blocks::Closed-Beta` item carries a `~priority::*` — both `~priority::1` * [x] PREP readiness assessment ([readiness!85](https://gitlab.com/gitlab-org/architecture/readiness/-/merge_requests/85)) out of draft — marked ready 2026-09-14, 180 days after opening, and split into ten team-scoped MRs. Reviewable, not merged * [ ] Beta user documentation merged — [!2271](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2271) still open at 13 days * [x] Metrics table filled in * [ ] Plan-table batch fill landed (S09, S16, monolith) — S09 landed 09-08 (20/20) and S16 09-11 (26/26), but the monolith fill [!2664](https://gitlab.com/gitlab-org/ops/artifact-registry/-/merge_requests/2664) is **open, not merged** * [ ] Anything not making Friday re-labelled `~AR-Blocks::GA` **explicitly**, not left to drift — #1149 and #1150 were de-labelled to **no** `~AR-Blocks::*` on 09-11 * [ ] Roadmap updated at docs/roadmap/closed-beta.md — last touched 2026-09-08 for the W36 close * [ ] Sprint summary posted to #f_artifact-registry Slack channel — the copy-paste text is in the W37 results note under `### Slack summary`; it has not been posted **Sprint summary for Slack** (copy-paste when closing): > Sprint W37 closed. Closed-beta blockers: {N} open, down from 16. S1 count: {N}. Readiness for 2026-09-11: {met / slipped, and why}. {One sentence on the biggest risk or win.} Sprint W38: {link}