Link agentic-era blog and release page to launch destinations

[cnunciato]

Add inline links to Joe's launch blog post for agent accounts, pulumi do, the npm package, the CoreWeave and NVIDIA providers, agent-friendly docs/CLI, skills, and the release page's blog roundup. Also adds an anchor to the release page's blog_section so the post can link to it.

[pulumi-bot]

Your site preview for commit bf0e6c1 is ready! 🎉

http://www-testing-pulumi-docs-origin-pr-19383-bf0e6c1c.s3-website.us-west-2.amazonaws.com

[github-actions]

Pre-merge Review — Last updated 2026-05-23T23:24:58Z

Tip

Summary: A small navigation-wiring PR (+11/-9 across two files) that drops seven inline links into existing paragraphs of the just-published Agentic Infrastructure Era launch blog, adds a [companion blog posts] link, plus a link: and anchor: from-the-blog to the matching release landing page so the post can deep-link back. No prose, examples, or factual claims are touched — every change is a wrapped phrase or a single frontmatter line. Reader-blocking wrongness would be a broken target or wrong anchor; I confirmed each of the seven internal destinations exists on disk and the #agent-friendly-markdown-docs-for-providers-and-components H2 actually matches a heading in the cited post. The trail's 17 unfaithful/unverifiable verdicts are all on pre-existing content the PR doesn't introduce, or are verifier mis-sources (npmjs.com Cloudflare bot-challenge 403; the verifier denying the existence of /releases/agentic-infrastructure-era/, which is literally the page being edited). Passes that ran: cited-claim spot-checks, frontmatter sweep, temporal-trigger sweep, editorial-balance pass; cross-sibling and code-execution skipped (not in a templated section; no programs touched).

Review confidence:

Dimension	Level	Notes
mechanics	HIGH	All seven new internal link targets exist on disk; #from-the-blog anchor is added in the same PR.
facts	HIGH	No factual claim is introduced or modified; all flagged claims pre-date this PR or are verifier mis-sources.

Investigation log

• Cross-sibling reads: not run (not in a templated section)
• External claim verification: 34 of 54 claims verified (6 unverifiable, 11 contradicted) · 4 specialists (numerical, cross-reference, capability, framing); 0 cross-specialist corroborations · routed: 0 inline, 19 Pass 1, 4 Pass 2 (verified 0, contradicted 2, unverifiable 2), 31 Pass 3 (verified 21, contradicted 7, unverifiable 3).
• Cited-claim spot-checks: 4 of 4 cited claims fetched and compared
• Frontmatter sweep: ran on body + meta_desc + social.{bluesky, linkedin, twitter}
• Temporal-trigger sweep: ran (recency words present in diff; spot-check in-review)
• Code execution: not run (no static/programs/ change)
• Code-examples checks: not run (no fenced code blocks in content files)
• Editorial-balance pass: ran (single-subject, N/A)

🚨 Outstanding	⚠️ Low-confidence	💡 Pre-existing	✅ Resolved
0	6	13	0

🔍 Verification trail

54 claims extracted · 34 verified · 6 unverifiable · 11 contradicted

• L33 in content/blog/the-agentic-infrastructure-era/index.md "The improvement from 33% to 86% on SWE-bench Verified represents a 4x reduction in the errors models make when coding." → ❌ contradicted (framing: shifted — the claim says "4x reduction in errors" but the actual arithmetic (67% → 14% error rate) yields ~4.8x, not 4x; the claim understates the improvement…; evidence: The math does not support "4x reduction in errors." Error rate drops from 67% (1−0.33) to 14% (1−0.86), a ratio of 0.67/0.14 ≈ 4.79x — closer to a 5x reduction, not 4x. A true 4x reduction would correspond to an ~83% success rate, not 86%.; source: WebSearch ran query "SWE-bench Verified 33% to 86% score improvement error reduction calculation"; arithmetic verified independently from the benchmark scores.)
• L33 in content/blog/the-agentic-infrastructure-era/index.md "Anthropic's Mythos model is not generally available at the time of this article (published 2026-05-19)." → ✅ verified (evidence: Anthropic's own Project Glasswing page states: "We do not plan to make Claude Mythos Preview generally available." As of the article's publication date (2026-05-19), the model remains restricted to a limited set of vetted partners via Proj…; source: https://www.anthropic.com/glasswing)
• L35 in content/blog/the-agentic-infrastructure-era/index.md "Anthropic's Claude Code is a coding-oriented agentic system." → ✅ verified (evidence: Anthropic's own product page states: "Claude Code is an agentic coding system that reads your codebase, makes changes across files, runs tests, and delivers committed code."; source: https://www.anthropic.com/product/claude-code)
• L39 in content/blog/the-agentic-infrastructure-era/index.md "Andrej Karpathy stated: 'Building a modern app is a bit like assembling IKEA furniture,' observing that writing the code was easy, fun, and fast, but the next…" → ❌ contradicted (framing: shifted — PR paraphrases the quote as being about what "the LLM wasn't naturally good at" and frames it as a "next mile" problem, while the source frames it as…; evidence: Karpathy's blog post (karpathy.bearblog.dev/vibe-coding-menugen/) says "Building a modern app is a bit like assembling IKEA future[sic]. There are all these services, docs, API keys, configurations, dev/prod deployments..." — the PR omits…; source: https://karpathy.bearblog.dev/vibe-coding-menugen/)
• L39 in content/blog/the-agentic-infrastructure-era/index.md "Andrej Karpathy made this observation nearly a year ago (relative to the article's publication date of May 2026)." → ✅ verified (framing: strengthened — the claim says "nearly a year ago" (relative to May 2026); the talk was June 2025, ~11 months prior, which fits the framing precisely.; evidence: Karpathy's widely-cited "Software 3.0" / "decade of agents" keynote was delivered at Y Combinator's AI Startup School on June 18–19, 2025 — approximately 11 months before the article's May 2026 publication date, consistent with "nearly a y…; source: WebSearch ran query "Andrej Karpathy 'Software 3.0' talk date 2025"; confirmed by multiple sources including singjupost.com (transcript dated June 18, 2025) and Apple Podcasts listing "19 JUNE 2025".)
• L41 in content/blog/the-agentic-infrastructure-era/index.md "At Pulumi, LLMs are now doing over 20% of infrastructure deployments, up from virtually zero a year ago." (also L177) → ✅ verified (evidence: The Pulumi blog post at pulumi.com/blog/the-agentic-infrastructure-era directly states: "LLMs are now doing over 20% of the infrastructure deployments, up from virtually zero a year ago."; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L41 in content/blog/the-agentic-infrastructure-era/index.md "LLM share of Pulumi infrastructure deployments was up from virtually zero a year ago." → ✅ verified (framing: strengthened — claim narrows the full source statement to only the "up from virtually zero a year ago" portion; source's broader form proves the claim as a sub…; evidence: The Pulumi blog post states: "LLMs are now doing over 20% of the infrastructure deployments, up from virtually zero a year ago." This directly confirms the claim's framing.; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L51 in content/blog/the-agentic-infrastructure-era/index.md "Pulumi supports infrastructure as code using programming languages including Python, TypeScript, Go, C#, Java, and more." → ✅ verified (framing: strengthened — claim lists a subset of supported languages (Python, TypeScript, Go, C#, Java) with "and more"; source's broader list proves the claim as a subs…; evidence: Pulumi's official docs confirm: "Pulumi's open source infrastructure as code SDK supports TypeScript, Python, Go, .NET, Java, and YAML." The claim's enumeration of Python, TypeScript, Go, C#, Java is a narrower subset of this broader list,…; source: https://www.pulumi.com/docs/iac/)
• L51 in content/blog/the-agentic-infrastructure-era/index.md "Pulumi began its journey with its open-source infrastructure-as-code project nearly ten years ago." → ❌ contradicted (framing: narrowed — claim broadens the timeline; the open-source project launched in 2018 (~7–8 years ago as of 2026), not "nearly ten years ago"; evidence: Pulumi open-sourced its IaC project on June 18, 2018 (per Pulumi's own blog: "One year ago today – on June 18, 2018 – we open sourced Pulumi"). As of 2026, that is approximately 7–8 years, not "nearly ten years." The company was founded in…; source: https://www.pulumi.com/blog/happy-birthday-to-pulumi-open-source/ and https://www.pulumi.com/about/)
• L55 in content/blog/the-agentic-infrastructure-era/index.md "Pulumi made the bet on infrastructure as code in programming languages nine years ago." → ✅ verified (framing: strengthened — the claim counts from the 2017 founding; the public open-source launch was June 2018 (~8 years ago from 2026), so "nine years" is slightly gener…; evidence: Pulumi was founded in 2017 and open-sourced on June 18, 2018. The blog post is written in 2026; counting from the 2017 founding gives ~9 years, consistent with the claim. Wikipedia confirms "Pulumi was founded in 2017 by former Microsoft e…; source: https://en.wikipedia.org/wiki/Pulumi)
• L107 in content/blog/the-agentic-infrastructure-era/index.md "Pulumi supports thousands of providers with a consistent programming model." → ❌ contradicted (framing: narrowed — claim broadens the actual figure ("thousands") vs. Pulumi's own stated "150+" or "180+" native providers; source supports the narrower form, not the…; evidence: Pulumi's own official docs and blog posts consistently cite 150+, 180+, or "nearly 200" native providers — not thousands. The August 2024 blog states "any of the 150+ cloud providers in the Pulumi Registry," and the 2024 Year in Review ref…; source: https://www.pulumi.com/blog/any-terraform-provider/ and https://www.pulumi.com/blog/pulumi-year-in-review/; intuition: The "thousands" figure is an order-of-magnitude overstatement relative to Pulumi's own published provider counts (~150–…)
• L117 in content/blog/the-agentic-infrastructure-era/index.md "A human can claim a Pulumi agent account at any time to make it permanent and to share with their team." → ✅ verified (evidence: The Pulumi blog post at pulumi.com/blog/the-agentic-infrastructure-era/ contains the verbatim text: "A human can claim that account at any time to make it permanent and to share with their team."; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L117 in content/blog/the-agentic-infrastructure-era/index.md "Agent accounts allow agents to use free, ephemeral Pulumi Cloud accounts straight from Claude Code, Codex, OpenCode, Copilot, Cursor, and others without a manu…" → ✅ verified (evidence: The agent-accounts docs page states: "Agent accounts give AI agents a free ephemeral Pulumi Cloud account to work in, without requiring a human user to sign up first." The blog post itself (confirmed via gh search) states: "Now agents can…; source: gh api repos/pulumi/docs/contents/content/docs/administration/organizations-teams/agent-accounts.md)
• L119 in content/blog/the-agentic-infrastructure-era/index.md "Pulumi now has an npm package that enables npx pulumi <anything>-style commands so agents can run any Pulumi command without needing to worry about installat…" → ❌ contradicted (framing: shifted — cited URL returns HTTP 403; broken citation must be fixed; evidence: The cited URL https://www.npmjs.com/package/pulumi returns HTTP 403 Forbidden with an empty body, making it a dead/inaccessible citation that cannot support the claim.; source: https://www.npmjs.com/package/pulumi)
• L121 in content/blog/the-agentic-infrastructure-era/index.md "pulumi do update cloudflare:r2:Bucket --tags '{ 'Foo': 'Bar' }' updates a bucket's tags." → ❌ contradicted (framing: shifted — the docs define the update operation as patch with <type> patch <id> --input-file, but the claim uses update <type> --tags with a different arg…; evidence: The official pulumi do docs show the update operation is called patch (not update), requires a provider resource ID, and uses --input-file for inputs: `pulumi do package:module:type patch --input-file <path…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
• L121 in content/blog/the-agentic-infrastructure-era/index.md "All thousands of providers and hundreds of thousands of resource types in the Pulumi ecosystem are supported by pulumi do across clouds." → 🤷 unverifiable (evidence: The blog post itself states "All thousands of providers and hundreds of thousands of resource types in the Pulumi ecosystem are supported across clouds." The "thousands of providers" figure is corroborated elsewhere in Pulumi docs, but the…; source: repo:content/blog/the-agentic-infrastructure-era/index.md; gh search code pulumi "thousands of providers"; intuition: The "hundreds of thousands of resource types" is a very large number — plausible if aggregating all Terraform-bridged p…)
• L121 in content/blog/the-agentic-infrastructure-era/index.md "The new pulumi do command enables direct create, read, update, delete, list, and API operations with a single command." → ✅ verified (framing: strengthened — the source says "create, read, update, delete, and query" plus "list"; the claim adds "API operations" as a characterization of provider functio…; evidence: The source blog post states: "With pulumi do, you can create, read, update, delete, and query any cloud resource from the terminal with a single command" and "Most resources support the full set of CRUD operations — create, read, update,…; source: content/blog/pulumi-do-direct-resource-operations/index.md)
• L123 in content/blog/the-agentic-infrastructure-era/index.md "Pulumi Cloud capabilities added to the CLI include change history, time-to-live stacks, drift detection, resource discovery and search, private registries, IDP…" → ✅ verified (evidence: The exact list of Pulumi Cloud capabilities added to the CLI is confirmed verbatim in the blog post: "things like change history, time-to-live stacks, drift detection, resource discovery and search, private registries, IDP, audit logs, sec…; source: repo:content/blog/the-agentic-infrastructure-era/index.md; gh search code --owner pulumi "time-to-live stacks")
• L123 in content/blog/the-agentic-infrastructure-era/index.md "The new Pulumi Cloud CLI additions include over 30 new commands." → ✅ verified (evidence: The blog post itself (and the GitHub search result) contains the exact phrase: "Over 30 new commands, you can think of this as the equivalent of the gh CLI which agents really like."; source: content/blog/the-agentic-infrastructure-era/index.md (confirmed via gh search code and read_file))
• L131 in content/blog/the-agentic-infrastructure-era/index.md "The pulumi neo CLI command runs the same agent that is already in Pulumi Cloud, sharing the same agentic loop." → ✅ verified (evidence: The blog post at content/blog/the-agentic-infrastructure-era/index.md states verbatim: "The new pulumi neo command lets you run the same agent that is already in Pulumi Cloud. It even shares the same agentic loop and uses a sophisticat…; source: repo:content/blog/the-agentic-infrastructure-era/index.md (confirmed via gh search code --owner pulumi "same agentic loop"))
• L131 in content/blog/the-agentic-infrastructure-era/index.md "The pulumi neo CLI uses a sophisticated architecture where the agent workstation is the user's local computer." → ➖ not-a-claim (evidence: The claim describes the PR author's (Pulumi/joe-duffy) own product architecture in their own blog post. The pulumi neo CLI is confirmed as a local TUI/agent command ("Run pulumi neo to start an interactive Neo session in your terminal"…; source: repo:content/blog/the-agentic-infrastructure-era/index.md; gh api repos/pulumi/docs/contents/content/docs/ai/_index.md)
• L133 in content/blog/the-agentic-infrastructure-era/index.md "Neo GitHub and Slack Apps allow users to @-mention Neo from GitHub pull requests and from Slack to kick off agentic infrastructure workflows." → ➖ not-a-claim (evidence: The claim describes Pulumi's own product feature (Neo GitHub and Slack Apps) in a Pulumi blog post authored by Pulumi's own team. This is a faithful description of the PR author's own design/pipeline — a first-party product announcement, n…; source: repo:content/blog/the-agentic-infrastructure-era/index.md)
• L135 in content/blog/the-agentic-infrastructure-era/index.md "The Neo Integration Catalog includes connectors to Atlassian, Datadog, Honeycomb, Linear, PagerDuty, and Supabase." → ✅ verified (evidence: The Pulumi Neo Integration Catalog blog post confirms: "Connect Neo to Atlassian, Datadog, Honeycomb, Linear, PagerDuty, and Supabase so your infrastructure agent has the context it needs to help." The same list also appears verbatim in th…; source: https://www.pulumi.com/blog/neo-integration-catalog/)
• L137 in content/blog/the-agentic-infrastructure-era/index.md "Neo Scheduled Tasks can automate recurring infrastructure tasks and can be confined to read-only operations." → ✅ verified (evidence: The source blog at /blog/neo-automations/ states "Permission mode is read-only, so a run can read state and propose changes through pull requests but can't apply changes directly," and the agentic-infrastructure-era blog (the file under re…; source: https://www.pulumi.com/blog/neo-automations/ and https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L145 in content/blog/the-agentic-infrastructure-era/index.md "The CoreWeave provider includes several examples and templates out of the box for training and inference architectures written in AI-native languages." → ✅ verified (evidence: The Pulumi blog post at pulumi.com/blog/the-agentic-infrastructure-era/ states verbatim: "includes several examples and templates out of the box for training and inference architectures written in AI-native languages."; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L145 in content/blog/the-agentic-infrastructure-era/index.md "The CoreWeave provider includes support for CoreWeave Kubernetes Service (CKS) and includes several examples and templates for training and inference architect…" → ❌ contradicted (framing: shifted — the claim attributes training/inference examples to the CoreWeave provider, but those examples belong to the NVIDIA AICR provider (pulumi-labs/pulumi…; evidence: The releases page confirms CKS support: "Includes support for all CoreWeave services, including CoreWeave Kubernetes Service (CKS)." However, the pulumi-coreweave repo's examples/ directory contains only a single basic subdirectory — the…; source: gh api repos/pulumi/pulumi-coreweave/contents/examples; gh api repos/pulumi/docs/contents/content/releases/agentic-infrastructure-era.md)
• L147 in content/blog/the-agentic-infrastructure-era/index.md "Using the CoreWeave provider and AICR provider together, users can spin up a CKS cluster and configure it with AICR in one program." → ✅ verified (evidence: The Pulumi blog post shows a Python code example using both pulumi_coreweave (to create a KubernetesCluster CKS cluster) and pulumi_nvidia_aicr (to configure it via aicr.ClusterStack) in a single program, directly confirming the cl…; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L147 in content/blog/the-agentic-infrastructure-era/index.md "The NVIDIA AI Cluster Runtime (AICR) provider delivers fully functional NVIDIA software stacks atop the underlying cloud provider's GPU infrastructure." → 🤷 unverifiable (evidence: The pre-fetched GitHub page returns HTTP 200 and the truncated body only shows the repo description: "Infrastructure as code for NVIDIA AI cluster runtime (AICR) in any language." This confirms the repo exists but does not contain enough c…; source: https://github.com/pulumi-labs/pulumi-nvidia-aicr)
• L153 in content/blog/the-agentic-infrastructure-era/index.md "Pulumi now serves its docs website in markdown to agents." → ✅ verified (evidence: The Pulumi blog post at pulumi.com/blog/the-agentic-infrastructure-era/ states: "We now serve our docs website in markdown to agents and did a pass over our documentation and CLI text to ensure it's maximally useful to agents."; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L155 in content/blog/the-agentic-infrastructure-era/index.md "Pulumi has added --json flags and structured errors across the CLI to help agents parse and react to outputs." → ✅ verified (evidence: The Pulumi blog directly states: "We have added --json and structured errors across the CLI to help agents parse and react to outputs appropriately." This matches the claim exactly.; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L157 in content/blog/the-agentic-infrastructure-era/index.md "Pulumi's new uber-skill describes how an agent should leverage capabilities from Level 0 (agent accounts), to Level 1 (pulumi do), to Level 2 (full IaC and `…" → ✅ verified (evidence: The blog file itself contains the exact passage: "We've done a pass over our skills and added a new uber-skill that describes how an agent should leverage the full suite of capabilities we now offer, from Level 0 (agent accounts), to Level…; source: gh search code --owner pulumi "uber-skill" → pulumi/docs:content/blog/the-agentic-infrastructure-era/index.md)
• L159 in content/blog/the-agentic-infrastructure-era/index.md "InfraBench is described as the infrastructure equivalent to SWE-bench." → ✅ verified (evidence: The Pulumi blog post states: "This is the infrastructure equivalent to SWE-bench and lets us measure stock agents, agents plus the new ..." — directly confirming InfraBench is described as the infrastructure equivalent to SWE-bench.; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L173 in content/blog/the-agentic-infrastructure-era/index.md "Compostable.ai is already living the 100% agentic infrastructure dream using Pulumi." → ✅ verified (evidence: The published Pulumi blog post at pulumi.com/blog/the-agentic-infrastructure-era states: "Compostable.ai already living the 100% agentic infrastructure dream" — an exact match to the claim. The Pulumi case study page further confirms Compo…; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L173 in content/blog/the-agentic-infrastructure-era/index.md "Wiz manages 1 million resources and does 100,000 daily deployments using Pulumi." → ✅ verified (framing: strengthened — the case study uses "over a million" and "hundreds of thousands of stack updates daily" (broader); the blog claim narrows to "1 million" and "10…; evidence: The Pulumi blog post itself states: "Wiz managing 1 million resources and doing 100k daily deployments." The Pulumi case study page corroborates with "over a million cloud resources" and "hundreds of thousands of infrastructure updates dai…; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/ (result index 2-6); https://www.pulumi.com/case-studies/wiz/ (result index 1-11,1-12))
• L173 in content/blog/the-agentic-infrastructure-era/index.md "A frontier lab grew their infrastructure footprint by 982% in a year using Pulumi." → ✅ verified (evidence: The Pulumi blog post at pulumi.com/blog/the-agentic-infrastructure-era/ states: "a frontier lab who grew their infrastructure footprint by 982% in a year, all using infrastructure modeled in true code."; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L175 in content/blog/the-agentic-infrastructure-era/index.md "Today's launches are detailed at /releases/agentic-infrastructure-era/ with companion blog posts." → 🤷 unverifiable (framing: shifted — claim asserts launches are detailed at /releases/agentic-infrastructure-era/ but all authoritative sources point to /blog/the-agentic-infrastructure-…; evidence: No evidence of a pulumi.com/releases/agentic-infrastructure-era/ page was found in any search result. Official Pulumi communications (PR Newswire, blog posts) consistently reference pulumi.com/blog/the-agentic-infrastructure-era as the…; source: WebSearch ran query "pulumi.com/releases/agentic-infrastructure-era/"; top results didn't surface the page — all official references point to /blog/the-agentic-infrastructure-era/; intuition: Pulumi's site uses /blog/ for launch announcements, not a /releases/ path; this URL pattern does not match any known Pu…)
• L175 in content/blog/the-agentic-infrastructure-era/index.md "Languages and verifiability already provide strong foundations, however, there are still clear areas we need to improve, beginning with making the full suite o…" → ❌ contradicted (framing: shifted — source_hint claims /releases/agentic-infrastructure-era/ but the actual URL is /blog/the-agentic-infrastructure-era/; evidence: The deep-dive content for the agentic infrastructure era launch lives at pulumi.com/blog/the-agentic-infrastructure-era/, not at /releases/agentic-infrastructure-era/. The PR Newswire press release explicitly states: "to dive deeper, p…; source: WebSearch ran query "pulumi.com/releases/agentic-infrastructure-era"; top results point to https://www.pulumi.com/blog/the-agentic-infrastructure-era/ and https://www.prnewswire.com/news-releases/pulumi-closes-the-ai-deployment-gap-with-agent-native-infrastructure-and-superintelligence-partnerships-302776114.html)
• L177 in content/blog/the-agentic-infrastructure-era/index.md "Pulumi states that LLMs currently handle over 20% of infrastructure deployments on its platform and the goal is to push toward 100%." → ❌ contradicted (framing: shifted — the source states the goal as "over 50% before the end of this year and well beyond afterwards"; the claim shifts this to "push toward 100%", which i…; evidence: The Pulumi blog confirms "LLMs are now doing over 20% of the infrastructure deployments," but the stated goal is "over 50% before the end of this year and well beyond afterwards" — not 100%. The 100% figure appears only in a third-party pa…; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L9 in content/releases/agentic-infrastructure-era.md "The agent surfaces depicted in the release hero image include Claude Code, OpenCode, Codex, and Pulumi Neo." (also L19) → ✅ verified (evidence: Both the feature_image_alt and hero_image_alt fields in the file read: "Agent surfaces — Claude Code, OpenCode, Codex, and Pulumi Neo — around the Pulumi mark", exactly matching the claim.; source: repo:content/releases/agentic-infrastructure-era.md)
• L24-27 in content/releases/agentic-infrastructure-era.md "Joe Duffy (co-founder and CEO of Pulumi) stated that LLMs are now doing over 20% of infrastructure deployments, up from virtually zero a year ago." → ❌ contradicted (framing: shifted — source uses collective "we at Pulumi" voice; claim attributes the statement specifically to "Joe Duffy (co-founder and CEO of Pulumi)"; evidence: The cited blog post does contain the statistic — "LLMs are now doing over 20% of the infrastructure deployments, up from virtually zero a year ago" — but it is written in first-person plural ("we're seeing something magical happen here at…; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)
• L44-47 in content/releases/agentic-infrastructure-era.md "A new Node.js package enables npx pulumi <anything>-style commands so that agents can discover and run Pulumi commands more easily." → ❌ contradicted (evidence: The cited URL https://www.npmjs.com/package/pulumi returns HTTP 403 Forbidden with an empty body, making it a dead/inaccessible citation that cannot support the claim.; source: https://www.npmjs.com/package/pulumi)
• L55-57 in content/releases/agentic-infrastructure-era.md "The new pulumi do command enables direct operations including create, read, update, delete, and list across all Pulumi-supported clouds and services." → ✅ verified (framing: strengthened — the source blog post also mentions "API operations" which the claim omits, but the claim's narrower subset is fully supported by the source.; evidence: The release file at content/releases/agentic-infrastructure-era.md (L55-57) reads: "The new pulumi do command enables direct operations like create, read, update, delete, and list across all Pulumi-supported clouds and services." The blo…; source: repo:content/releases/agentic-infrastructure-era.md; repo:content/blog/the-agentic-infrastructure-era/index.md)
• L63-65 in content/releases/agentic-infrastructure-era.md "Ephemeral Pulumi Cloud accounts (agent accounts) allow agents like Claude Code and Codex to spin up Pulumi-managed infrastructure without requiring a human in…" → ➖ not-a-claim (evidence: The text at L63-65 is the PR author's own description of the "Agent accounts" feature card in the release page: "Ephemeral Pulumi Cloud accounts allow agents like Claude Code and Codex to spin up Pulumi-managed infrastructure without requi…; source: repo:content/releases/agentic-infrastructure-era.md)
• L73-75 in content/releases/agentic-infrastructure-era.md "Pulumi Cloud in the CLI adds dozens of new commands for Pulumi operations previously only available in the browser." → ✅ verified (evidence: The releases file itself (L73-75 card) reads: "Dozens of new commands for Pulumi operations previously only available in the browser." The linked blog post corroborates: "Up to now, most of what you could do with Pulumi Cloud had to be don…; source: repo:content/releases/agentic-infrastructure-era.md and repo:content/blog/better-cli-interactions-for-agents-and-humans/index.md)
• L74 in content/releases/agentic-infrastructure-era.md "The new Pulumi Cloud CLI is compared to the gh CLI (GitHub CLI) in terms of its role and agent usability." → ✅ verified (evidence: Line ~74 of the file contains the card "Pulumi Cloud in the CLI" with the description: "Think of it like the gh CLI (which agents love) for Pulumi Cloud." — directly comparing the new Pulumi Cloud CLI to the GitHub CLI (gh) in terms of…; source: repo:content/releases/agentic-infrastructure-era.md)
• L81-82 in content/releases/agentic-infrastructure-era.md "Neo is described as Pulumi's infrastructure agent that previously existed only in the Pulumi Cloud console and is now being expanded to other surfaces." → ✅ verified (evidence: Lines 81-82 of the file read: "With this release, Neo, our infrastructure agent, moves out of the Pulumi Cloud console and into more of the places where the work of managing infrastructure happens." This directly confirms Neo is Pulumi's i…; source: repo:content/releases/agentic-infrastructure-era.md)
• L90-92 in content/releases/agentic-infrastructure-era.md "The new pulumi neo command brings Neo out of Pulumi Cloud and into the terminal, enabling agentic infrastructure anywhere Pulumi can run." → ✅ verified (framing: strengthened — claim slightly paraphrases the source ("your terminal" → "the terminal", "so you can do agentic infrastructure anywhere you can run Pulumi" → "e…; evidence: The release file itself (L90-92) reads: "The new pulumi neo command brings Neo out of Pulumi Cloud and into your terminal, so you can do agentic infrastructure anywhere you can run Pulumi." The blog post at /blog/pulumi-neo-cli/ confirms…; source: repo:content/releases/agentic-infrastructure-era.md and repo:content/blog/pulumi-neo-cli/index.md)
• L99-102 in content/releases/agentic-infrastructure-era.md "Neo can be @-mentioned in GitHub issues and pull requests, and in Slack workspaces, to kick off infrastructure tasks." → ✅ verified (framing: strengthened — claim says "GitHub issues and pull requests" and "Slack workspaces" to "kick off infrastructure tasks"; source confirms both surfaces and uses "…; evidence: The official Pulumi blog at /blog/neo-github-slack/ states: "Mention @pulumi-neo in a pull request or issue and Neo replies in the thread. Mention @neo in a Slack channel and Neo starts a task." The agentic infrastructure era blog also con…; source: https://www.pulumi.com/blog/neo-github-slack/)
• L108-110 in content/releases/agentic-infrastructure-era.md "Neo's new integration catalog includes connectors for Atlassian, Datadog, Honeycomb, Linear, PagerDuty, and Supabase." → ✅ verified (evidence: The Pulumi blog post at pulumi.com/blog/neo-integration-catalog/ states: "Connect Neo to Atlassian, Datadog, Honeycomb, Linear, PagerDuty, and Supabase so your infrastructure agent has the context it needs to help." The agentic-infrastruct…; source: https://www.pulumi.com/blog/neo-integration-catalog/)
• L130-133 in content/releases/agentic-infrastructure-era.md "CoreWeave is described as 'the leader in GPU infrastructure'." → 🤷 unverifiable (framing: shifted — CoreWeave's own positioning uses "industry-leading GPU performance" and "AI-native cloud," not "leader in GPU infrastructure"; the claim's framing is…; evidence: No source uses the exact phrase "leader in GPU infrastructure" to describe CoreWeave. CoreWeave's own site calls itself "purpose-built to deliver industry-leading GPU performance" and "one of the most performant, reliable, and resilient AI…; source: WebSearch ran query "CoreWeave 'leader in GPU infrastructure'"; top results didn't contain that exact phrase from any authoritative source.)
• L131-134 in content/releases/agentic-infrastructure-era.md "The new CoreWeave provider includes support for all CoreWeave services, including CoreWeave Kubernetes Service (CKS)." → ✅ verified (evidence: The file content/releases/agentic-infrastructure-era.md contains the exact text: "Includes support for all CoreWeave services, including CoreWeave Kubernetes Service (CKS)." The pulumi/pulumi-coreweave SDK confirms CKS support with `Ck…; source: repo:content/releases/agentic-infrastructure-era.md; gh search code --owner pulumi CKS --repo pulumi/pulumi-coreweave; intuition: The phrase "all CoreWeave services" is a broad marketing claim that cannot be exhaustively verified against CoreWeave's…)
• L139-143 in content/releases/agentic-infrastructure-era.md "The NVIDIA AI Cluster Runtime (AICR) provider delivers fully functional NVIDIA software stacks atop the underlying cloud provider's GPU infrastructure." → 🤷 unverifiable (evidence: The cited URL returns HTTP 200 and the truncated body shows only the repo title: "Infrastructure as code for NVIDIA AI cluster runtime (AICR) in any language." The body is too truncated to confirm or deny the specific claim that the provid…; source: https://github.com/pulumi-labs/pulumi-nvidia-aicr)
• L152-155 in content/releases/agentic-infrastructure-era.md "Pulumi now serves all of its documentation, including the complete Pulumi Registry, in agent-friendly Markdown." → 🤷 unverifiable (framing: narrowed — the Registry API evidence supports Markdown for provider/component docs specifically; the claim's "all documentation" scope is broader and unconfirm…; evidence: Registry pages confirm the Pulumi Cloud Registry API serves docs with Accept: text/markdown support for provider/component packages, but the claim that Pulumi serves all of its documentation (including the complete Registry) in agent-f…; source: WebSearch ran query "Pulumi agent-friendly markdown docs providers Registry 2025"; top results show Registry API markdown support but don't address the "all documentation" scope of the claim or the specific blog post at /blog/better-cli-interactions-for-agents-and-humans/)
• L181-183 in content/releases/agentic-infrastructure-era.md "InfraBench is described as the infrastructure equivalent of SWE-bench." → ✅ verified (evidence: The Pulumi blog post states: "This is the infrastructure equivalent to SWE-bench and lets us measure stock agents, agents plus the new..." — directly confirming InfraBench is described as the infrastructure equivalent of SWE-bench.; source: https://www.pulumi.com/blog/the-agentic-infrastructure-era/)

📊 Editorial balance

Single-subject post; balance check N/A.

🚨 Outstanding in this PR

These must be resolved or refuted before merging.

No outstanding issues.

⚠️ Low-confidence

Review each and resolve as appropriate — these don't block the PR.

Style findings

Found by pattern-based linting; Findings may be false positives.

• line 117: [style] wordiness — 'benefit from' is too wordy.
• line 119: [style] wordiness — 'subsequent' is too wordy.
• line 121: [style] wordiness — 'it is' is too wordy.
• line 121: [style] difficulty qualifier — Avoid difficulty qualifier 'easy' -- it judges difficulty for the reader (STYLE-GUIDE.md §Inclusive Language).
• line 145: [style] wordiness — 'all of' is too wordy.
• line 145: [style] weasel word — 'several' is a weasel word!

📋 Triaged verifier findings

I double-checked these and realized they weren't real findings — click to expand

• [L119] content/blog/the-agentic-infrastructure-era/index.md — "npm package that enables npx pulumi <anything>-style commands…" — Mis-sourced: npmjs.com returns HTTP 403 cf-mitigated: challenge to any non-browser request — that's Cloudflare's bot-protection challenge, not a dead URL. The page loads fine for human readers and is the canonical npm package page; the verifier saw the challenge response and concluded the citation was broken. (Reproduced with a real-browser User-Agent: still 403 + cf-mitigated: challenge.)

• [L44-47] content/releases/agentic-infrastructure-era.md — link: https://www.npmjs.com/package/pulumi added to the one-command-execution card. — Mis-sourced: Same Cloudflare bot-challenge 403 as L119 — the npmjs.com package page exists and is correct; the verifier's fetcher is blocked at the edge.

• [L175] content/blog/the-agentic-infrastructure-era/index.md — "/releases/agentic-infrastructure-era/ is the wrong URL for the launch deep-dive." — Mis-sourced: That path is a real page in this repo — content/releases/agentic-infrastructure-era.md (the very file the PR also edits). .frontmatter-validation.json shows it rendering at /releases/agentic-infrastructure-era/. The verifier WebSearched the public web, didn't surface a freshly-cut release page, and inferred the URL was wrong; the page is just new.

• [L175] content/blog/the-agentic-infrastructure-era/index.md — "Today's launches are detailed at /releases/agentic-infrastructure-era/." — Mis-sourced: Same as above. The PR adds anchor: from-the-blog to that page's blog_section in the same commit, so /releases/agentic-infrastructure-era/#from-the-blog resolves once this PR ships.

💡 Pre-existing issues in touched files (optional)

These factual concerns predate this PR — the only change to each line was adding a Markdown link to the bold lead-in or wrapping an existing phrase. Flagging them here for future-author awareness; they should not block this PR.

• [L33] content/blog/the-agentic-infrastructure-era/index.md — "The improvement from 33% to 86% on SWE-bench Verified represents a 4x reduction in the errors models make when coding." — Pre-existing: Line 33 is outside any hunk in this PR's diff (first change starts at L114). Verifier's arithmetic check (67% → 14% error rate ≈ 4.8x) suggests "5x reduction" would be more precise; flagging for the original blog author, not this linker.

• [L39] content/blog/the-agentic-infrastructure-era/index.md — Karpathy "IKEA furniture" framing. — Pre-existing: Line 39 is well outside any diff hunk. The framing concern (paraphrasing what the LLM "wasn't naturally good at" vs. the source's "all these services, docs, API keys" framing) is about original blog prose this PR does not touch.

• [L51] content/blog/the-agentic-infrastructure-era/index.md — "Pulumi began its journey … nearly ten years ago." — Pre-existing: Line 51, not in any diff hunk. (Founded 2017, OSS launch 2018 ⇒ ~8 years as of 2026.) Original-author concern.

• [L107] content/blog/the-agentic-infrastructure-era/index.md — "We support thousands of providers with a consistent programming model." — Pre-existing: Line 107, not in any diff hunk. Pulumi's own published figures are ~150–200 native providers (more once Terraform-bridged providers are counted); this is an original-author wording concern.

• [L121] content/blog/the-agentic-infrastructure-era/index.md — "pulumi do update cloudflare:r2:Bucket --tags '{ "Foo": "Bar" }' updates a bucket's tags." — Pre-existing: This PR's only change on L121 was wrapping the bold lead-in in [Imperative infrastructure operations](/blog/pulumi-do-direct-resource-operations/). — the example syntax was already in the blog. If the docs really standardized on patch instead of update, that's a fix for the blog author (and arguably for pulumi-do-direct-resource-operations/index.md), not this linker.

• [L121] content/blog/the-agentic-infrastructure-era/index.md — "All thousands of providers and hundreds of thousands of resource types in the Pulumi ecosystem are supported across clouds." — Pre-existing: Same paragraph as above — the marketing-scale figures are original blog prose, not introduced by this PR.

• [L145] content/blog/the-agentic-infrastructure-era/index.md — "…includes several examples and templates out of the box for training and inference architectures written in AI-native languages." — Pre-existing: This PR's only change on L145 was wrapping [CoreWeave provider](/registry/packages/coreweave/). — the "examples and templates" wording predates the PR. If those examples belong to the AICR provider rather than the CoreWeave provider, that's a fix for the blog author.

• [L147] content/blog/the-agentic-infrastructure-era/index.md — NVIDIA AICR "fully functional NVIDIA software stacks" description. — Pre-existing: This PR wrapped the bold lead-in in [NVIDIA AI Cluster Runtime (AICR) provider](https://github.com/pulumi-labs/pulumi-nvidia-aicr).; the descriptive sentence predates the PR. Verifier marked it unverifiable because the GitHub repo README was truncated — not a blocker.

• [L177] content/blog/the-agentic-infrastructure-era/index.md — "…push this beyond 20% and toward 100% in the years ahead." — Pre-existing: Line 177 is outside any diff hunk this PR introduces. The "100% vs. 50%" goal-framing concern is original-author prose.

• [L24-27] content/releases/agentic-infrastructure-era.md — "Joe Duffy, co-founder and CEO" attribution on the hero quote. — Pre-existing: Lines 24–27 are not in any diff hunk (this PR's release-file changes are at L47 and ~L185). The release page is allowed to attribute a paraphrased version of the blog quote to its author; not introduced here.

• [L130-133] content/releases/agentic-infrastructure-era.md — CoreWeave described as "the leader in GPU infrastructure." — Pre-existing: Lines 130–133, not in any diff hunk. Marketing-language concern for the release-page author.

• [L139-143] content/releases/agentic-infrastructure-era.md — NVIDIA AICR "fully functional NVIDIA software stacks atop the underlying cloud provider's GPU infrastructure." — Pre-existing: Same wording as the L147 blog claim, on lines this PR does not touch. AICR repo README is sparse, but that's an upstream-doc issue, not a PR blocker.

• [L152-155] content/releases/agentic-infrastructure-era.md — "Pulumi now serves all of its documentation … in agent-friendly Markdown." — Pre-existing: Lines 152–155, not in any diff hunk. "All documentation" vs. provider/component-docs-only scope is for the release-page author.

✅ Resolved since last review

No items resolved since the last review.

📜 Review history

• 2026-05-23T23:24:58Z — Link-wiring PR (+11/-9); all seven internal targets exist on disk; the trail's 11 contradicted + 6 unverifiable verdicts triage to mis-sourced (npmjs Cloudflare 403; verifier denying the existence of the very /releases/agentic-infrastructure-era/ page being edited) or pre-existing on lines this PR did not touch. (bf0e6c1)

---

Need a re-review? Want to dispute a finding? Mention @claude and include #update-review.
(For ad-hoc questions or fixes, just @claude — no hashtag.)

[sicarul]

LGTM!