Decision needed: approval path for workprentice-authored content PRs (4 PRs, 12-27 days open)

[workprentice]

Four content PRs have been open 12-27 days, all CI-green, all reviewed clean by the automated pre-merge check, all labelled review:no-blockers, none stale. Each was explicitly handed to a named human reviewer who has not yet acted. Two more PRs are parked on a separate, related question. I'm not asking for a merge here — I will not self-merge any of these — I'm asking for a routing decision so this stops being a recurring nudge cycle.

Decision 1 — who is the final approver for net-new editorial/marketing content, and on what turnaround?

PR	Title	Open since	Handoff	Current state
#20515	Two Real-World Incidents: DIY State Backend risk	27 days	CamSoper tagged @cnunciato 2026-08-18 for final review; requested byline fix pushed same day	CI green, review:no-blockers, no response from @cnunciato
#20536	Best Terraform Cloud Alternative	24 days	Handed to @cnunciato by CamSoper 2026-08-14; nudged 2026-08-18	CI green, review:no-blockers, no response from @cnunciato
#20609	Refresh: Best IaC Tools for 2026	21 days	Handed to @cnunciato by CamSoper 2026-08-14; nudged 2026-08-19	CI green, review:no-blockers, no response from @cnunciato
#20779	Why AI Agents Write Better Infra in Real Languages	12 days	No human has engaged since it opened; flagged to @CamSoper 2026-08-18	CI green, review:no-blockers, no response

Every one of these has been reviewer-approved-clean by our automated pre-merge check well before the nudge dates above, so the gap isn't content quality, it's an assigned reviewer with no committed turnaround.

Options, so this is a quick decision rather than an open-ended one:

• (a) @cnunciato remains the approver for this category, with an agreed SLA (e.g. review within N business days of handoff).
• (b) @CamSoper approves and merges this category directly instead of routing to @cnunciato.
• (c) content that clears the automated pre-merge review cleanly auto-merges after N days with no objection raised.

Decision 2 — SEO-flavored content ownership

#20756 (CDK vs Terraform vs Pulumi) and #20881 (What Is AI Infrastructure? pillar page) are both explicitly on standby per CamSoper's 2026-08-18 comments, pending a decision on who owns SEO-content review now that there's no SEO czar. I'm not pushing on either of these further; I said I'd stand by and I am. Flagging the open question here so it doesn't get lost: I can carry the SEO review workload if that's useful (that's squarely in my job description), but the final approval call stays with a human either way.

What this issue is and isn't

This is a routing/ownership question, not a request to merge anything. I won't self-merge any PR in this queue. Happy to make further content changes on any of these if something comes up in review.

---

🧠 This issue was created by workprentice.

[CamSoper]

Right call to make this one decision instead of six nudges. Before it gets answered, the queue is larger than the six PRs listed — there are ten open workprentice PRs, and the four missing ones fall squarely inside the two decisions rather than outside them.

Belongs in Decision 1 (net-new editorial content, CI-green, review:no-blockers, not stale):

PR	Title	Open since	Handoff
#20949	Platform Engineering in the Agentic Era	3 days	@CamSoper tagged @cnunciato 2026-08-18: "Tagging in @cnunciato for approval and merging"
#20631	GPU Kubernetes infrastructure for AI workloads	20 days	No human has engaged since it opened — the same state as #20779

#20949 matters most to get in front of the decision: it was handed to @cnunciato on exactly the same day and in exactly the same words as #20515, so whatever turnaround gets agreed in Decision 1 should cover it. Leaving it out means answering the question and still having an unrouted PR in the same category the next morning.

Worth knowing while deciding: #20949 and #20779 are both bylined joe-duffy. That's a different risk shape from the three pulumi-content-team posts. Option (c) — auto-merge on a clean automated review after N days — reads very differently for a post going out under the founder's name than for a house-bylined comparison piece. If (c) is attractive for the rest, it may still be worth carving those two out.

Arguably Decision 2, worth a yes or no: #20917 and #20919, the language-list normalization pair. They're SEO-flavored but they're fixes rather than net-new content, so they may be deliberately out of scope here — if so, say so explicitly, because they're currently sitting in the same "handed over, no response" state as everything else and will otherwise be the next thing that needs a nudge.

That makes the real shape 10 PRs: 6 net-new editorial, 2 SEO standby, 2 SEO fixes. Answering for all ten closes the loop; answering for six leaves four to re-litigate.

---

Generated by Claude Code

[cnunciato]

@CamSoper Thanks for the nudge. I'll take a look at these. These came at me a little from the side and I didn't have any context, so haven't prioritized them. But will make some decisions on them today.

[CamSoper]

@cnunciato -- Gotta be honest here, that entire comment was Claude acting on my behalf while triaging PRs. Didn't mean to nudge quite so... impersonally. Sorry about that!

[workprentice]

Pulling together everything I found this pass so this is one decision-ready update instead of another round of nudges.

Where things stand

30 content PRs are open, totaling about 1,908 PR-days. Since the last update: two old sprint PRs and one hold-breached PR left the queue via merge, one closed unmerged. Nothing else moved on its own — the queue is still accumulating roughly 30 PR-days per day it sits.

I spent this pass fixing what I could fix myself and diagnosing the rest, rather than merging anything or pinging people individually.

Group 1 — clean, needs only a yes/no

These have no outstanding review finding and no unmet request from anyone. All were explicitly handed to @cnunciod for final approval between 8/14 and 8/18:

• Add: Best Terraform Cloud Alternative for Large Multi-Cloud Teams blog post #20536 — Terraform Cloud alternative comparison page
• blog: GPU Kubernetes infrastructure for AI workloads #20631 — GPU Kubernetes infrastructure for AI workloads
• CDK vs. Terraform vs. Pulumi: new three-way comparison page #20756 — CDK vs. Terraform vs. Pulumi three-way comparison (CamSoper routed this to an SEO-content-ownership question rather than a content one — see the note below)
• Add What Is AI Infrastructure? pillar page #20881 — "What Is AI Infrastructure?" pillar page (same SEO-ownership question)
• fix(seo): normalize Pulumi language lists on gads landing pages, templates, and brand-consistency surfaces #20919 — SEO language-list normalization on the Google Ads landing pages
• Add: Platform Engineering in the Agentic Era #20949 — Platform Engineering in the Agentic Era
• blog: Why AI Agents Write Better Infrastructure in Real Languages than HCL #20779 — AI agents vs. HCL comparison post (I just closed out its last blocking finding, see below)
• Refresh: Best Infrastructure as Code (IaC) Tools for 2026 — tables, versions, quotes, FAQs #20609 — IaC tools 2026 refresh (currently shows review:stale again after a rebase; a fresh @claude #update-review will clear that, no content work needed)
• Add self-serve trial lane to /request-a-demo/ #21063 / Add self-routing qualifier lines to /contact/ (Deltas 2 & 3) #21067 — the /request-a-demo/ and /contact/ funnel changes flagged in the parent card, still green and unreviewed

#20779 and #20949 carry a joe-duffy byline, per CamSoper's flag on 8/21 — worth excluding from any blanket timer-based auto-merge policy, since a byline post reads differently than routine content and probably wants an explicit look regardless of how long it's been clean.

Group 2 — needs named-author work, not mine to do

• blog: Two Real-World Incidents That Expose the Risk of a DIY State Backend #20515 (mine) — I fixed the three CI lint failures (title over 60 characters, feature image at the wrong size and background color) and confirmed the byline fix already landed. The fresh review this triggered found one real, pre-existing issue from an earlier round: the lede's rewrite dropped a qualifier that now has the post contradicting itself about self-managed-backend locking (state-and-backends.md:228 vs. this post's L32). That one needs a content fix before merge — I've left it for the next round rather than editing prose beyond what the CI failure required.
• [blog] Launch post: Terraform modules in the Pulumi Cloud registry #20648 (fnune) — the review's one blocking item was a merge-order dependency on [docs] Document Terraform modules in the Pulumi Cloud registry #19453, which merged three weeks ago. I confirmed the target page is live and triggered a re-check; it surfaced a second, separate issue — the post's canonical_url points at the docs reference page instead of itself, which would deindex the announcement. fnune needs to point canonical_url at the post itself.
• Make self-hosted Pulumi Cloud a first-class story #20914 (joeduffy) — the pinned review has specific, concrete findings on the new ECS self-hosting config reference: it names three config keys the installer doesn't actually read (route53ZoneName, route53Subdomain, openSearchUser), two required keys it omits (recaptchaSiteKey, recaptchaSecretKey), and marks the AG Grid license key as required when the installer treats it as optional. All five are checkable against pulumi-self-hosted-installers source, which the review already did.

Group 3 — the orphaned "content sprint" block (16 PRs, sicarul, ~100 days old)

I refreshed the pinned review on all 16. Two came back clean; the other 14 show one or more findings. Reading a representative sample closely, two different things are happening and they call for different handling:

1. Pure decay, content already sound. [CONTENT SPRINT] Snowflake Warehouses, Roles, and Short-Lived Credentials with Pulumi #18952 (Snowflake creds post) is the clean case: every one of its 18 previously-unverified provider claims checked out this round with zero content corrections needed. What's actually stale is mechanical — a committed meta.png that now conflicts with the site's newer build-time social-card generator, and a publish date that's already passed. That's a one-line-per-item fix, not a rewrite.
2. Real regressions from a rebase. [CONTENT SPRINT] Crossplane to Pulumi: A Migration Map for Platform Teams #18963 and [CONTENT SPRINT] From CloudFormation to Pulumi Without a Rewrite #18966 show history rewritten since last review — the branches were force-pushed at some point, and fixes the May reviews had already recorded as resolved are back: [CONTENT SPRINT] Crossplane to Pulumi: A Migration Map for Platform Teams #18963's KCL "imperative" wording error is back, and [CONTENT SPRINT] From CloudFormation to Pulumi Without a Rewrite #18966's Custom Resources factual claim is back. These need the author to redo work that was already done once.

My read: #18961 and #18965 are ready to land now (clean review, no findings). #18952 and likely more of the batch are a quick refresh — decay bumps, not rewrites — but that needs sicarul (or someone standing in for a closed-out author) to actually push the bump commits; I can't push to someone else's PR. #18963 and #18966 need real re-work, not just a bump. I haven't individually read all 14 remaining findings, so I'd treat the rest case-by-case using this same lens rather than assume they're all one type.

One item worth flagging regardless of review outcome: #18964 (Neo Slash Commands) describes a fast-moving product surface. A hundred days is a long time for Neo's actual command set — I'd want that checked against the current product before it ships, independent of what the automated review found.

These 16 are sicarul's PRs to close out or hand off, not mine — I'm surfacing the split so whoever picks this block up isn't treating all 16 the same way.

The actual ask

You said on 8/21 these came at you from the side without context and you hadn't prioritized them. Hopefully the groupings above give you that context now — Group 1 needs only a decision, Group 2 needs a specific named-author fix each, Group 3 needs someone (sicarul or a delegate) to work through 16 refreshed reviews with the decay-vs-regression split I found.

Two ways to unblock the yes/no group specifically: name yourself as approver with a turnaround window, or delegate content-PR approval to CamSoper for anything that isn't a byline post. Either works from where I sit; what doesn't work is the current state, where "handed off" and "decided" look identical from outside.

To be explicit: I'm not going to merge any of these myself. Given what happened with #20882, that's not a case-by-case judgment call anymore — it's a hard rule for this queue.