Add pulumi do blog post, docs page, and agent accounts by arunkumar611 · Pull Request #19291 · pulumi/docs

arunkumar611 · 2026-05-20T20:05:07Z

Summary

Blog post introducing pulumi do (direct resource operations) for the May 22 launch. Covers provider functions (shipped in v3.242.0) and resource CRUDL operations (create, read, patch, delete, list). Includes Karpathy framing, agent account integration, and roadmap (ESC, cross-resource references, stateful mode).
Docs reference page at /docs/iac/cli/direct-resource-operations/ with full flag reference, command syntax for both functions and resources, and comparison table.
Placeholder docs page for agent accounts at /docs/administration/organizations-teams/agent-accounts/. Engineering team to update with implementation details before launch.

Resolves https://github.com/pulumi/pulumi-service/issues/43128.

🤖 Generated with Claude Code

Blog post introducing `pulumi do` for the May 22 launch, covering provider functions and resource CRUDL operations. Docs reference page with full flag reference, command syntax, and examples. Placeholder docs page for agent accounts under organizations-teams. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

github-actions · 2026-05-20T20:12:42Z

Social Media Review

content/blog/pulumi-do-direct-resource-operations/index.md

No social copy was provided for any platform. Suggested copy drafted below from the blog body.

X — missing

LinkedIn — missing

Bluesky — missing

Suggested copy

X (231/255 chars) — drafted from the article:

Spinning up a bucket with IaC means: make a directory, create a project, configure credentials, write the code, preview, deploy. Several steps for a single resource.

We built pulumi do to collapse all of that. Here's what it does.

LinkedIn (1284/2950 chars) — drafted from the article:

When you need a quick bucket, a lookup on an existing VPC, or a one-off database, infrastructure as code can be more overhead than the operation deserves. Making a directory, creating a project, configuring credentials, writing the code, previewing, deploying — that's the minimal path through IaC for a resource you might delete in an hour.

Today we're shipping pulumi do, a new Pulumi CLI command for direct resource operations.

It runs create, read, update, delete, list, and query operations against any resource from thousands of Pulumi providers — the same providers that power full Pulumi programs — with no project, no code, and no state file required. AWS, Azure, Google Cloud, Kubernetes, Cloudflare, Datadog, and every other provider in the registry.

We designed it for humans and AI coding agents equally: consistent command structure across every provider, JSON on stdout, progress on stderr, consistent exit codes. An agent that learns the pattern for one provider knows it for all of them.

When a one-off resource eventually needs to grow into something permanent, there's a path. The roadmap includes a stateful mode that accumulates resource state and ejects to a full Pulumi program when the time comes.

Ships today as a research preview in Pulumi CLI v3.242.0.

Bluesky (272/300 chars) — drafted from the article:

Creating one S3 bucket with IaC means making a directory, configuring a project, writing code, and deploying. That overhead is why we built pulumi do.

Direct resource operations — create, read, update, delete — across thousands of providers, no project or state required.

To apply these suggestions, comment: @claude please update the social posts in the frontmatter with the suggested copy from the social media review above

To re-run the social media review after updates, comment: /social-review

Updated for commit d435ed1808d7d8977ccd3b26969ccc18d86e74a8 (short: d435ed1) at 2026-05-23 01:50 UTC.

pulumi-bot · 2026-05-20T20:14:49Z

Your site preview for commit d435ed1 is ready! 🎉

http://www-testing-pulumi-docs-origin-pr-19291-d435ed18.s3-website.us-west-2.amazonaws.com

github-actions · 2026-05-20T20:20:09Z

Pre-merge Review — Last updated 2026-05-20T22:30:22Z

Tip

Summary: Launch package for pulumi do and agent accounts — a new blog post announcement, a CLI reference page under content/docs/iac/cli/ (paralleling api.md, exit-codes.md, environment-variables.md), and an administration page paralleling accounts.md / teams.md. All four previously-outstanding issues were addressed in 60f9989: the Karpathy quote was removed, the Datadog claim was softened to accurately describe scope, the agent-accounts meta description was corrected to remove "deployments", and the eight-steps enumeration was completed with eight distinct steps. No new findings introduced by the fix push.

Review confidence:

Dimension	Level	Notes
mechanics	HIGH
facts	MEDIUM	Several claims about brand-new features (preview-state behavior, planned ESC integration, claim-window timing) can't be confirmed against public sources yet — see ⚠️ Low-confidence.
cross-sibling consistency	HIGH
code correctness	HIGH

Investigation log

Cross-sibling reads: 5 of 8 siblings (accounts.md, teams.md for the docs page; api.md, environment-variables.md, exit-codes.md for the CLI page)
External claim verification: 75 of 125 claims verified (22 unverifiable, 7 contradicted) · 4 specialists (numerical, cross-reference, capability, framing); 0 cross-specialist corroborations · routed: 0 inline, 97 Pass 1, 7 Pass 2 (verified 2, contradicted 2, unverifiable 3), 21 Pass 3 (verified 4, contradicted 3, unverifiable 14).
Cited-claim spot-checks: 7 of 7 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: ran (3 specialists: structural, existence, body-code-coverage); 0 findings
Editorial-balance pass: ran (single-subject, N/A)

🚨 Outstanding	⚠️ Low-confidence	💡 Pre-existing	✅ Resolved
0	24	0	4

🔍 Verification trail

125 claims extracted · 75 verified · 22 unverifiable · 7 contradicted

L3 in content/blog/pulumi-do-direct-resource-operations/index.md "date: 2026-05-22T06:00:00-07:00" → ➖ not-a-claim (evidence: The "date" front-matter field in a blog post is metadata set by the PR author for their own content's publication date — it is a scheduling/authoring decision, not a falsifiable assertion about an external fact. A future-dated timestamp in…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md L3)
L5 in content/blog/pulumi-do-direct-resource-operations/index.md "meta_desc: 'pulumi do lets you create, read, update, delete, and query any cloud resource from the terminal with a single command. No IaC project required.'" → 🤷 unverifiable (evidence: No public documentation or authoritative source for a pulumi do command was found. The official Pulumi CLI commands page lists standard commands (up, stack, config, destroy, etc.) but makes no mention of pulumi do or a "direct resource…; source: WebSearch ran query "pulumi do CLI command CRUD cloud resource no IaC project"; top results didn't address the claim; intuition: The pulumi do command described in this blog post does not appear in any current public Pulumi CLI documentation — it…)
L31 in content/blog/pulumi-do-direct-resource-operations/index.md "The bucket exists. The result is JSON on stdout. No project, no stack, no state file." → ➖ not-a-claim (evidence: The sentence "The bucket exists. The result is JSON on stdout. No project, no stack, no state file." is the PR author's own description of the behavior of the direct resource operations feature being introduced in the blog post — it is a f…; source: WebSearch ran query "Pulumi direct resource operations blog 2025 no project no stack no state"; claim is author's own design description, not a third-party attribution.)
L41 in content/blog/pulumi-do-direct-resource-operations/index.md "Joe Duffy described the agentic infrastructure era and framed the key design principle that agents should start with the smallest possible commitment and progr…" → 🤷 unverifiable (evidence: The blog post at pulumi.com/blog/the-agentic-infrastructure-era is confirmed to exist and is associated with Joe Duffy (Pulumi CEO), covering the "agentic infrastructure era" concept. However, the specific design principle that "agents sho…; source: WebSearch ran query "pulumi.com/blog/the-agentic-infrastructure-era 'smallest possible commitment' 'progressively'"; top results confirmed the post exists but didn't surface the specific design principle framing.)
L41 in content/blog/pulumi-do-direct-resource-operations/index.md "Pulumi shipped a redesigned CLI so both humans and agents can find the right command on the first try." → 🤷 unverifiable (framing: shifted — sources describe CLI improvements (JSON output, agent-friendly flags, 30+ Cloud features added to CLI) but do not confirm "redesigned CLI" or "find t…; evidence: A Pulumi blog post titled "Better CLI Interactions for Agents and Humans" exists and is prominently featured, and sources describe CLI improvements like JSON output, structured errors, and agent-friendly flags. However, the specific framin…; source: WebSearch ran query "Pulumi 'better CLI interactions for agents and humans' blog redesigned"; top results show the blog title exists but don't surface the specific claim's framing.)
L43 in content/blog/pulumi-do-direct-resource-operations/index.md "pulumi do ships with provider functions (read-only queries against any cloud API) and resource operations (create, read, patch, delete, list) across all 150+…" → ✅ verified (evidence: Pulumi's own docs confirm "now over 150 providers and 7,500 resource types" in the Pulumi Registry (from pulumi/docs content/blog/2025-product-launches/index.md and pulumi-release-notes-117/index.md), directly supporting the "150+ Pulumi p…; source: gh search code --owner pulumi "150 providers" → pulumi/docs:content/blog/2025-product-launches/index.md and content/blog/pulumi-release-notes-117/index.md)
L51-53 in content/blog/pulumi-do-direct-resource-operations/index.md "Andrej Karpathy wrote: 'I didn't even spend all that much work in the code editor itself. I spent most of it in the browser, moving between tabs and settings a…" → ❌ contradicted (framing: shifted — the quoted text is not present in the fetched page body; the attribution may be to a different post or the quote may be fabricated/misattributed; evidence: The cited URL returns HTTP 200 and contains the blog post, but the exact quoted passage — "I didn't even spend all that much work in the code editor itself. I spent most of it in the browser, moving between tabs and settings and configurin…; source: https://karpathy.bearblog.dev/vibe-coding-menugen/)
L51-53 in content/blog/pulumi-do-direct-resource-operations/index.md "Andrej Karpathy wrote: 'I didn't even spend all that much work in the code editor itself. I spent most of it in the browser, moving between tabs and settings a…" → ❌ contradicted (framing: shifted — the quoted text does not appear in the cited source; the source discusses vibe coding MenuGen but does not contain the attributed quote; evidence: The pre-fetched page does not contain the quoted passage "I didn't even spend all that much work in the code editor itself. I spent most of it in the browser, moving between tabs and settings and configuring and gluing a monster." The body…; source: https://karpathy.bearblog.dev/vibe-coding-menugen/)
L55 in content/blog/pulumi-do-direct-resource-operations/index.md "Account creation, credential plumbing, configuration files, service wiring. LLMs can write the code, but the infrastructure around it is where they stall. We'r…" → 🤷 unverifiable (evidence: verification did not converge within 8 turns)
L57 in content/blog/pulumi-do-direct-resource-operations/index.md "The straightest-line path to creating a Postgres database on AWS with IaC is eight steps: create a project, pick a language, install packages, write infrastruc…" → 🤷 unverifiable (evidence: The claim asserts "eight steps" but only enumerates six (create a project, pick a language, install packages, write infrastructure code, preview, deploy). No authoritative Pulumi source confirms a canonical eight-step process for creating…; source: WebSearch ran query "Pulumi new project steps create deploy AWS postgres"; top results didn't address an eight-step process.; intuition: Claim says "eight steps" but only names six — the number is likely wrong or the enumeration is incomplete; reviewer sho…)
L59 in content/blog/pulumi-do-direct-resource-operations/index.md "pulumi do collapses the IaC workflow into one command with no project directory, no Pulumi.yaml, and no state backend." (also L193) → ✅ verified (evidence: The blog post states verbatim: "pulumi do collapses those eight steps into one. No project directory, no Pulumi.yaml, no state backend." This exactly matches the claim's framing.; source: content/blog/pulumi-do-direct-resource-operations/index.md)
L63 in content/blog/pulumi-do-direct-resource-operations/index.md "pulumi do dynamically builds its CLI tree from any installed Pulumi provider's schema at runtime." → ✅ verified (evidence: The blog post at L63 states "The command dynamically builds its CLI tree from any installed Pulumi provider's schema. Every resource type and every function becomes a subcommand at runtime." The source code in pulumi/pulumi at `pkg/cmd/p…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md and gh api repos/pulumi/pulumi/contents/pkg/cmd/pulumi/do/do.go)
L63 in content/blog/pulumi-do-direct-resource-operations/index.md "pulumi do provides the same type checking and default handling that a full Pulumi program provides, applied at the CLI layer before anything touches the clou…" → ✅ verified (evidence: The blog post at the referenced file contains the exact sentence: "You get the same type checking and default handling that a full Pulumi program provides, applied at the CLI layer before anything touches the cloud." — this is a verbatim m…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L79 in content/blog/pulumi-do-direct-resource-operations/index.md "Provider functions shipped in Pulumi CLI v3.242.0." → 🤷 unverifiable (evidence: The cited URL returns HTTP 200, but the body is severely truncated and contains no usable content beyond the page header — there is no release notes text to confirm or deny that "provider functions" shipped in v3.242.0.; source: https://github.com/pulumi/pulumi/releases/tag/v3.242.0)
L103 in content/blog/pulumi-do-direct-resource-operations/index.md "The bound PCL in pulumi do runs through the same evaluation logic the engine uses for resource registrations (defaults, type casting, secret marking)." → ➖ not-a-claim (evidence: The claim is a verbatim quote from the blog post itself (authored by fraser-waters, pat-gavlin, arun-loganathan): "The bound PCL then runs through the same evaluation logic the engine uses for resource registrations (defaults, type casting…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L103 in content/blog/pulumi-do-direct-resource-operations/index.md "The provider's Invoke call executes and the result lands as JSON on stdout." → ➖ not-a-claim (evidence: The sentence "The provider's Invoke call executes, and the result lands as JSON on stdout" is the PR author's own description of the internal pipeline of the pulumi do feature being introduced in this very blog post — it is a faithful…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L105 in content/blog/pulumi-do-direct-resource-operations/index.md "YAML input files work with pulumi do when --input yaml is passed alongside --input-file; the CLI converts the YAML through a converter plugin before eval…" → ➖ not-a-claim (evidence: The claim is a faithful description of the PR author's own design/pipeline as documented in the blog post itself: "YAML input files also work. Pass --input yaml alongside --input-file and the CLI converts the YAML through a converter p…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L111 in content/blog/pulumi-do-direct-resource-operations/index.md "Secrets in function results appear as [secret] by default in pulumi do output." → ✅ verified (evidence: The blog file at L111 states: "Secrets in function results appear as [secret] by default. Pass --show-secrets to reveal them." This is confirmed by the Pulumi source code in pkg/cmd/pulumi/do/do_function.go which returns "[secret]"…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md (L111) and gh search: pulumi/pulumi:pkg/cmd/pulumi/do/do_function.go)
L111 in content/blog/pulumi-do-direct-resource-operations/index.md "The --dry-run flag in pulumi do sets the operation to preview mode, signaling the provider to return placeholder values instead of live data." → ✅ verified (evidence: The blog post at content/blog/pulumi-do-direct-resource-operations/index.md contains the exact sentence: "The --dry-run flag sets the operation to preview mode, signaling the provider to return placeholder values instead of live data." —…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L115 in content/blog/pulumi-do-direct-resource-operations/index.md "Each pulumi do resource operation maps to a provider CRUD method, using the same provider logic a full Pulumi program would use." → ➖ not-a-claim (evidence: The text at the relevant line in the blog post reads: "Each operation maps to a provider CRUD method, using the same provider logic a full Pulumi program would use." This is the PR author's own description of their own feature design in th…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L129 in content/blog/pulumi-do-direct-resource-operations/index.md "The pulumi do CLI prompts for confirmation before creating a resource; pass --yes to auto-approve." → ✅ verified (evidence: The blog post itself states at the relevant section: "The CLI prompts for confirmation before creating. Pass --yes to auto-approve." This exactly matches the claim.; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L129 in content/blog/pulumi-do-direct-resource-operations/index.md "Passing --yes to pulumi do create auto-approves the creation without prompting." → ✅ verified (evidence: The blog post explicitly states: "The CLI prompts for confirmation before creating. Pass --yes to auto-approve." and later reinforces: "All mutating operations (create, patch, delete) prompt for confirmation. Pass --yes to skip the pro…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L129-139 in content/blog/pulumi-do-direct-resource-operations/index.md "On successful pulumi do create, the output is JSON with the cloud provider ID and all resource properties." → ✅ verified (evidence: The blog post explicitly states: "On success, the output is JSON with the cloud provider ID and all resource properties" and shows a JSON example with "id" and "properties" fields for a successful pulumi do aws s3 Bucket create comma…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L148 in content/blog/pulumi-do-direct-resource-operations/index.md "pulumi do read output follows the {'id', 'properties'} shape." → ➖ not-a-claim (evidence: The line "The output follows the same {"id", "properties"} shape." is the PR author's own description of the pulumi do feature's output contract being introduced in this blog post — it is a faithful description of the author's own design…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L150-151 in content/blog/pulumi-do-direct-resource-operations/index.md "The pulumi do patch operation reads the current state, merges the provided changes, shows a diff, and prompts for confirmation." → ✅ verified (evidence: The blog post itself states: "Patch (update) a resource by providing new inputs. The CLI reads the current state, merges your changes, shows a diff, and prompts for confirmation" — an exact match to the claim.; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L156 in content/blog/pulumi-do-direct-resource-operations/index.md "The pulumi do delete operation prompts for confirmation before destroying a resource." → ✅ verified (evidence: The blog post explicitly states under the Delete operation: "Delete a resource. The CLI prompts for confirmation before destroying" and later reinforces: "All mutating operations (create, patch, delete) prompt for confirmation."; source: content/blog/pulumi-do-direct-resource-operations/index.md)
L168 in content/blog/pulumi-do-direct-resource-operations/index.md "The --all flag in pulumi do list retrieves every resource; --count N limits the result set." → 🤷 unverifiable (evidence: verification did not converge within 8 turns)
L168 in content/blog/pulumi-do-direct-resource-operations/index.md "The output of pulumi do <type> list is a JSON array of {'id', 'name'} objects." → ➖ not-a-claim (framing: strengthened — the claim is a faithful description of the PR author's own design/pipeline for the pulumi do list command being introduced in this blog post.; evidence: The claim describes the PR author's own design of the pulumi do feature. The blog post at L168 states "The output is a JSON array of {\"id\", \"name\"} objects" — this is the author's own description of the behavior of a new feature be…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L170 in content/blog/pulumi-do-direct-resource-operations/index.md "All mutating operations (create, patch, delete) in pulumi do prompt for confirmation; passing --yes skips the prompt." → ✅ verified (evidence: The blog post at L170 states verbatim: "All mutating operations (create, patch, delete) prompt for confirmation. Pass --yes to skip the prompt, which is useful for scripting and agent workflows." This exactly matches the claim.; source: content/blog/pulumi-do-direct-resource-operations/index.md)
L174 in content/blog/pulumi-do-direct-resource-operations/index.md "pulumi do resolves provider configuration through ambient credentials (environment variables and credential files already present in the shell, like AWS_ACC…" → ➖ not-a-claim (evidence: The claim is a direct quote from the blog post itself (the PR author's own description of their feature): "Today, pulumi do` resolves provider configuration through ambient credentials (environment variables and credential files already p…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L185 in content/blog/pulumi-do-direct-resource-operations/index.md "An agent running Claude Code, Codex, or Cursor can go from zero (no Pulumi account, no CLI installed) to creating cloud resources in a single session using `pu…" → 🤷 unverifiable (evidence: verification did not converge within 8 turns)
L189 in content/blog/pulumi-do-direct-resource-operations/index.md "Consistent command structure across every provider. The <package> <module> <type> <operation> pattern is the same for AWS, Azure, Google Cloud, Kubernete…" → 🤷 unverifiable (evidence: The cited source /blog/better-cli-interactions-for-agents-and-humans/ was not surfaced in search results, and no other authoritative Pulumi source was found confirming the specific <package> <module> <type> <operation> CLI command patt…; source: WebSearch ran query "Pulumi better CLI interactions blog consistent command structure agents"; top results didn't address the claim or surface the cited blog post.)
L193 in content/blog/pulumi-do-direct-resource-operations/index.md "Datadog's CLI is CI-only." → ❌ contradicted (framing: shifted — the claim asserts the CLI is "CI-only" but sources show it is a general-purpose CLI installable and usable locally, with CI/CD pipelines being one op…; evidence: Datadog's datadog-ci CLI can be installed globally via npm/yarn and run locally outside CI. The Serverless docs state "You can use the CLI to instrument your AWS Lambda functions" and "You can also add the command to your CI/CD pipelines"…; source: https://docs.datadoghq.com/serverless/libraries_integrations/cli/ and https://github.com/datadog/datadog-ci)
L193 in content/blog/pulumi-do-direct-resource-operations/index.md "Cloudflare's CLI covers only a subset of its API." → ✅ verified (framing: strengthened — claim narrows the source's broader statement ("many products aren't available across every interface") to the specific CLI/API subset relationsh…; evidence: Cloudflare's own blog states "Many Cloudflare products have no CLI commands in Wrangler," and they are rebuilding the CLI specifically because "today, many of our products aren't available across every one of these interfaces" — directly c…; source: https://blog.cloudflare.com/cf-cli-local-explorer/)
L193 in content/blog/pulumi-do-direct-resource-operations/index.md "pulumi do supports 150+ providers under one syntax." → ✅ verified (evidence: The blog post itself states "resource operations (create, read, patch, delete, list) across all 150+ Pulumi providers" and "the syntax is the same across all 150+ providers in the Pulumi ecosystem." Multiple other authoritative Pulumi docs…; source: content/blog/pulumi-do-direct-resource-operations/index.md (lines ~43, ~75); corroborated by pulumi/docs:layouts/index.llms.txt and pulumi/docs:content/what-is/what-is-pulumi.md)
L203 in content/blog/pulumi-do-direct-resource-operations/index.md "We're building Pulumi ESC integration into pulumi do so you manage credentials once and resolve them everywhere:" → ✅ verified (evidence: The /docs/esc/ path resolves to the Pulumi ESC documentation home page, which describes "Pulumi ESC (Environments, Secrets, and Configuration)" — confirming the link target is valid and the product name matches the claim.; source: repo:content/docs/esc/_index.md)
L211 in content/blog/pulumi-do-direct-resource-operations/index.md "The planned --env flag for pulumi do will reference an ESC environment by name, and ESC handles credential resolution including OIDC-based dynamic credenti…" → 🤷 unverifiable (evidence: verification did not converge within 8 turns)
L213 in content/blog/pulumi-do-direct-resource-operations/index.md "Pulumi ESC provides one credential layer across every provider, with rotation, RBAC, and audit built in." → ✅ verified (framing: strengthened — the source describes these as separate features; the claim summarizes them as a unified "one credential layer" with "rotation, RBAC, and audit b…; evidence: The Pulumi ESC concepts page confirms centralized credential management across providers, with "Role-Based Access Control (RBAC) — Control who can read, write, or delete environments" and "Audit logs — Track all access and modifications to…; source: repo:content/docs/esc/concepts/_index.md)
L236 in content/blog/pulumi-do-direct-resource-operations/index.md "When resources are projected into a full IaC program with pulumi convert, the generated code contains subnet.vpcId = myVpc.id rather than a hard-coded stri…" → 🤷 unverifiable (evidence: verification did not converge within 8 turns)
L244 in content/blog/pulumi-do-direct-resource-operations/index.md "In the planned stateful mode, the first pulumi do mutation will silently create a default project at $PULUMI_HOME/do/default/ with a dev stack." → ✅ verified (framing: strengthened — source says "automatically creates"; claim says "silently create" — both convey the same automatic/unprompted behavior; the source's broader for…; evidence: The docs page content/docs/iac/cli/direct-resource-operations.md under "Stateful mode (planned)" states: "On first mutation, pulumi do automatically creates a default project at $PULUMI_HOME/do/default/ with a dev stack." The claim…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L248 in content/blog/pulumi-do-direct-resource-operations/index.md "The pulumi init --import-from-do command is planned to generate a Pulumi project in the user's chosen language with all resources imported and dependency gra…" → 🤷 unverifiable (evidence: verification did not converge within 8 turns)
L250 in content/blog/pulumi-do-direct-resource-operations/index.md "Resources created via pulumi do can be governed by Pulumi Insights from day one, even before opting into a full IaC workflow." → ➖ not-a-claim (evidence: The claim is the PR author's own description of how their new pulumi do feature integrates with Pulumi Insights — a faithful description of the author's own product design. The /product/ page confirms Pulumi Insights provides multi-clo…; source: content/product/_index.md)
L252 in content/blog/pulumi-do-direct-resource-operations/index.md "pulumi do uses the same providers, resource types, and property schemas as pulumi up." → ➖ not-a-claim (evidence: The claim is a description of the PR author's own design for pulumi do, stated within the same blog post: "Each operation maps to a provider CRUD method, using the same provider logic a full Pulumi program would use" and "You get the sam…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)
L261 in content/blog/pulumi-do-direct-resource-operations/index.md "The documentation for pulumi do is at /docs/iac/cli/direct-resource-operations/." → 🤷 unverifiable (evidence: The pre-fetched URL (https://github.com/pulumi/pulumi/releases/tag/v3.242.0) returned HTTP 200 but the body is truncated to just the navigation header, providing no content about documentation paths for pulumi do. The claim about the doc…; source: https://github.com/pulumi/pulumi/releases/tag/v3.242.0)
L265 in content/blog/pulumi-do-direct-resource-operations/index.md "- Documentation" → 🤷 unverifiable (evidence: verification did not converge within 8 turns)
L266 in content/blog/pulumi-do-direct-resource-operations/index.md "- File a feature request" → ✅ verified (evidence: The URL https://github.com/pulumi/pulumi/issues/new returns HTTP 200, confirming the link is live and points to the GitHub new issue page for the pulumi/pulumi repository, which is the correct destination for filing a feature request.; source: https://github.com/pulumi/pulumi/issues/new)
L267 in content/blog/pulumi-do-direct-resource-operations/index.md "- Pulumi Community Slack for discussion" → ✅ verified (evidence: The URL https://slack.pulumi.com/ returns HTTP 200 and confirms it is the Pulumi Community Slack: "Join the Pulumi Community on Slack where you can ask questions or share ideas about infrastructure as code."; source: https://slack.pulumi.com/)
L3 in content/docs/administration/organizations-teams/agent-accounts.md "Agent accounts let AI agents provision ephemeral Pulumi Cloud accounts automatically so agents can use state, ESC, and deployments without manual signup." → ❌ contradicted (framing: narrowed — claim broadens the source by adding "deployments" to the list of supported capabilities; the document only confirms state and ESC, not deployments; evidence: The file's own body ("What agents can do with the account") lists only "Store and manage infrastructure state across sessions" and "Use Pulumi ESC for secrets and configuration" — deployments are never listed as a capability of agent accou…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L15 in content/docs/administration/organizations-teams/agent-accounts.md "Agent accounts are in preview status, and behavior and limits may change based on feedback." → ✅ verified (evidence: The file at L15 contains the notes shortcode: "Agent accounts are in preview. Behavior and limits may change based on feedback." — which directly matches the claim's assertion about preview status and potential changes based on feedbac…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L18 in content/docs/administration/organizations-teams/agent-accounts.md "When the Pulumi CLI detects it is running in an agent context and no Pulumi Cloud credentials are available, it creates an account silently and continues." → ➖ not-a-claim (evidence: The text at line 18 of agent-accounts.md is the PR author's own description of the agent accounts feature they are introducing: "When the Pulumi CLI detects it is running in an agent context and no Pulumi Cloud credentials are available,…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L18 in content/docs/administration/organizations-teams/agent-accounts.md "The human claims the agent-provisioned account later to take permanent ownership." → ✅ verified (framing: strengthened — claim adds "agent-provisioned" qualifier to "account"; source's broader form proves the claim as a subset of the described behavior.; evidence: Line 18 of the file states: "The human claims the account later to take permanent ownership." The claim adds "agent-provisioned" as a descriptor, which is consistent with the surrounding context describing agent-provisioned accounts.; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L22 in content/docs/administration/organizations-teams/agent-accounts.md "AI agents that can trigger agent account creation include Claude Code, Codex, and Cursor, or any tool that calls the Pulumi CLI." → ✅ verified (framing: strengthened — claim narrows the broader Pulumi announcement (any agentic coding tool calling the CLI) to three named examples plus the general case; source's…; evidence: Pulumi's press release names "Agentic coding tools like Claude Code, OpenAI Codex, Cursor" as the tools in context, and The New Stack confirms Pulumi is rolling out "free, ephemeral Pulumi Cloud accounts that an agent can spin up on its ow…; source: https://www.morningstar.com/news/pr-newswire/20260519sf62492/pulumi-closes-the-ai-deployment-gap-with-agent-native-infrastructure-and-superintelligence-partnerships; https://thenewstack.io/pulumi-infrastructure-agent-era/)
L23 in content/docs/administration/organizations-teams/agent-accounts.md "The Pulumi CLI detects there are no Pulumi Cloud credentials by checking for the absence of a PULUMI_ACCESS_TOKEN environment variable and no entry in ~/.pu…" → ✅ verified (framing: strengthened — the doc describes this as step 2 of the "How it works" flow; the claim accurately narrows it to the credential-detection mechanism specifically; evidence: The doc at L23 states verbatim: "The CLI detects there are no Pulumi Cloud credentials (no PULUMI_ACCESS_TOKEN, no entry in ~~/.pulumi/credentials.json)." The two credential sources (PULUMI_ACCESS_TOKENenv var and~~/.pulumi/credenti…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md (L23); gh api repos/pulumi/pulumi/contents/pkg/workspace/creds.go)
L24 in content/docs/administration/organizations-teams/agent-accounts.md "Instead of prompting for login, the Pulumi CLI provisions a new Pulumi Cloud account and prints a claim link on stderr." → ✅ verified (evidence: The file at line 24 (step 3 of "How it works") reads verbatim: "Instead of prompting, the CLI provisions a new Pulumi Cloud account and prints a claim link on stderr." This is an exact match to the claim.; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L25-26 in content/docs/administration/organizations-teams/agent-accounts.md "The user can claim the agent account at any time by visiting the claim link and signing in." → ✅ verified (evidence: The file at line 25-26 states: "The user claims the account at any time by visiting the link and signing in." This directly matches the claim's assertion that the user can claim the agent account at any time by visiting the claim link and…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L30 in content/docs/administration/organizations-teams/agent-accounts.md "The agent account claim window is 30 days from creation." → ❌ contradicted (framing: shifted — claim states "30 days from creation" but the source reports the expiry/claim window is 72 hours from creation, not 30 days.; evidence: The New Stack (published 1 day ago, covering Pulumi's agent account launch) states: "An agent-registered account expires after 72 hours, but a human can claim it later to make it permanent." The PR claims the window is 30 days, not 72 hour…; source: https://thenewstack.io/pulumi-infrastructure-agent-era/)
L30 in content/docs/administration/organizations-teams/agent-accounts.md "The agent account claim window is 30 days." → ❌ contradicted (framing: shifted — source describes a 72-hour account expiry (not a 30-day claim window); the two figures refer to different durations for the same feature; evidence: A recent (May 2026) report on Pulumi's agent account feature states: "An agent-registered account expires after 72 hours, but a human can claim it later to make it permanent." No source mentions a 30-day claim window.; source: WebSearch ran query "Pulumi agent account claim window days"; top result: https://thenewstack.io/pulumi-infrastructure-agent-era/; intuition: 30 days is a suspiciously round number for an ephemeral agent account; 72 hours (3 days) is the figure consistently cit…)
L31 in content/docs/administration/organizations-teams/agent-accounts.md "https://app.pulumi.com/claim/abc123" → 🤷 unverifiable (evidence: The URL https://app.pulumi.com/claim/abc123 returns HTTP 200, but the body is a generic Pulumi Cloud SPA shell (CSS/font definitions only) with no content specific to "abc123" — it appears to be a placeholder/example URL that renders the a…; source: https://app.pulumi.com/claim/abc123; intuition: The URL path "/claim/abc123" looks like a placeholder example URL (with "abc123" as a dummy token), not a real claimabl…)
L34 in content/docs/administration/organizations-teams/agent-accounts.md "The agent does not block on the claim and continues running commands while the user decides whether to claim." → ✅ verified (evidence: The file at content/docs/administration/organizations-teams/agent-accounts.md explicitly states: "The agent does not block on the claim. It continues running commands while the user decides whether to claim."; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L34 in content/docs/administration/organizations-teams/agent-accounts.md "Only operations that require a claimed identity — specifically inviting teammates and billing — wait for the claim." → ✅ verified (evidence: The file at the relevant line reads: "Only operations that require a claimed identity (inviting teammates, billing) wait for the claim." This matches the claim exactly, with only a minor stylistic difference (em-dash vs. parentheses).; source: content/docs/administration/organizations-teams/agent-accounts.md)
L38 in content/docs/administration/organizations-teams/agent-accounts.md "The provisioned agent account is a Pulumi Cloud individual account." → ✅ verified (evidence: The file at line 38 (under "What agents can do with the account") states verbatim: "The provisioned account is a Pulumi Cloud individual account."; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L40 in content/docs/administration/organizations-teams/agent-accounts.md "Agent accounts can store and manage infrastructure state across sessions." → ✅ verified (evidence: The file at content/docs/administration/organizations-teams/agent-accounts.md under "What agents can do with the account" explicitly lists: "Store and manage infrastructure state across sessions" as a bullet point describing agent accoun…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L41 in content/docs/administration/organizations-teams/agent-accounts.md "Agent accounts can use Pulumi ESC for secrets and configuration." → ✅ verified (evidence: The file at the relevant section explicitly lists "Use Pulumi ESC for secrets and configuration" as one of the things agents can do with the provisioned account. The ESC docs confirm Pulumi ESC provides "centralized secrets m…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L45 in content/docs/administration/organizations-teams/agent-accounts.md "Claiming an agent account replaces the placeholder identity with the user's real identity, and all state, history, and resources transfer automatically with no…" → 🤷 unverifiable (evidence: No public Pulumi documentation page for "agent accounts" (/docs/administration/organizations-teams/agent-accounts) was found. No source describes a "claim" mechanism that replaces a placeholder identity and automatically transfers state,…; source: WebSearch ran query "Pulumi Cloud 'agent account' claim placeholder identity organizations teams"; top results didn't address the claim; intuition: The described "agent account claim" feature — automatic no-migration transfer of all state, history, and resources — do…)
L47 in content/docs/administration/organizations-teams/agent-accounts.md "Claiming an agent account never reduces capability; everything available before the claim is still available after, plus features gated on having a real identi…" → 🤷 unverifiable (evidence: External sources confirm Pulumi agent accounts are ephemeral and "a human can claim it later to make it permanent," but no public documentation addresses whether claiming never reduces capability or specifically unlocks inviting teammates,…; source: WebSearch ran query "Pulumi agent account claim never reduces identity features"; top results confirm ephemeral account/claim mechanic (thenewstack.io) but don't address the capability-preservation guarantee.; intuition: The "never reduces capability" guarantee is a strong, absolute claim about a brand-new feature; it may be correct by de…)
L51 in content/docs/administration/organizations-teams/agent-accounts.md "The agent account active period is 3 days (72 hours) from creation, during which the account is fully functional." → ✅ verified (framing: strengthened — source says account "expires after 72 hours"; claim frames this as a 3-day active period during which the account is "fully functional." The sou…; evidence: A recent (May 2026) The New Stack article covering Pulumi's agent account feature states: "An agent-registered account expires after 72 hours, but a human can claim it later to make it permanent." This confirms the 3-day (72-hour) active p…; source: https://thenewstack.io/pulumi-infrastructure-agent-era/)
L53 in content/docs/administration/organizations-teams/agent-accounts.md "Claiming an agent account after the active period (but within the 30-day claim window) unlocks the account and restores full functionality." → 🤷 unverifiable (evidence: The New Stack reports that "an agent-registered account expires after 72 hours, but a human can claim it later to make it permanent," but no public source specifies a 30-day claim window or the detail about restoring full functionality aft…; source: WebSearch ran query "Pulumi agent accounts ephemeral claim 30 days window restore functionality"; top results didn't address the specific 30-day claim window figure.; intuition: The "30-day claim window" is a specific numerical claim on a brand-new feature (agent accounts launched ~May 2026); no…)
L55 in content/docs/administration/organizations-teams/agent-accounts.md "The user's cloud resources are never touched if the agent account locks permanently." → 🤷 unverifiable (evidence: No authoritative Pulumi documentation was found describing a "permanent lock" state for agent accounts or confirming that cloud resources are never touched in such a scenario. Search results confirm agent accounts are ephemeral (expire aft…; source: WebSearch ran query "Pulumi agent accounts permanent lock cloud resources" and "Pulumi agent accounts ephemeral permanently locked cloud resources never touched"; top results didn't address the claim.; intuition: The claim's framing ("permanently locked" account → cloud resources "never touched") is a strong safety guarantee that…)
L57 in content/docs/administration/organizations-teams/agent-accounts.md "Cloud resources (such as S3 buckets, VMs, and databases) created by the agent continue running in the user's cloud provider account regardless of the Pulumi Cl…" → ✅ verified (framing: strengthened — claim uses "the user's cloud provider account" where source says "your cloud provider account"; semantically equivalent in context; evidence: The file at line ~57 states: "Cloud resources (S3 buckets, VMs, databases) created by the agent continue running in your cloud provider account regardless of the Pulumi Cloud account's state." This is an exact match to the claim (with only…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L61 in content/docs/administration/organizations-teams/agent-accounts.md "The CLI does not provision a new account if:" → ➖ not-a-claim (evidence: Line 61 reads "The CLI does not provision a new account if:" — this is a section heading in the PR author's own documentation describing their own CLI's designed behavior, not a falsifiable assertion attributed to a third party.; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L63-65 in content/docs/administration/organizations-teams/agent-accounts.md "The Pulumi CLI does not provision a new agent account if a valid PULUMI_ACCESS_TOKEN environment variable is set." → ✅ verified (evidence: The file at content/docs/administration/organizations-teams/agent-accounts.md states under "When agent accounts are not created": "The CLI does not provision a new account if: A valid PULUMI_ACCESS_TOKEN environment variable is set" —…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L67 in content/docs/administration/organizations-teams/agent-accounts.md "If multiple agents run on the same machine (for example, Cursor and Claude Code), the second agent reuses the credentials the first agent created." → ✅ verified (evidence: The file content/docs/administration/organizations-teams/agent-accounts.md contains the exact sentence verbatim in the "When agent accounts are not created" section: "If multiple agents run on the same machine (for example, Cursor and Cl…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L71 in content/docs/administration/organizations-teams/agent-accounts.md "The automatic agent account provisioning only triggers when the CLI detects it is running under an AI agent; humans still see the standard login prompt." → ✅ verified (evidence: The file at L71 (Limitations section) states verbatim: "Agent context only. The automatic provisioning only triggers when the CLI detects it is running under an AI agent. Humans still see the standard login prompt." This is an exact ma…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L72 in content/docs/administration/organizations-teams/agent-accounts.md "Agent accounts support only one owner; multi-admin is not available until the account is claimed and upgraded." → ✅ verified (evidence: The file at L72 (Limitations section) states verbatim: "Single user. Agent accounts support one owner. Multi-admin is not available until the account is claimed and upgraded." This exactly matches the claim.; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L73 in content/docs/administration/organizations-teams/agent-accounts.md "After 72 hours (3 days), unclaimed agent accounts go read-only, and the user must claim within 30 days to restore full access." → 🤷 unverifiable (evidence: Public sources confirm "an agent-registered account expires after 72 hours, but a human can claim it later to make it permanent" (The New Stack, May 2026), but no source corroborates the specific "read-only" intermediate state after 72 hou…; source: WebSearch ran query "Pulumi agent accounts unclaimed expiry hours days read-only"; top result: https://thenewstack.io/pulumi-infrastructure-agent-era/; intuition: The "30 days to restore full access" figure and the "read-only" intermediate state are suspiciously specific details no…)
L74 in content/docs/administration/organizations-teams/agent-accounts.md "If the claim link is lost and no email was set, recovery requires running the CLI on the same machine where the account was created." → ➖ not-a-claim (evidence: The text at L74 reads verbatim: "If the claim link is lost and no email was set, recovery requires running the CLI on the same machine where the account was created." This is a description of the PR author's own design for the Agent Accoun…; source: repo:content/docs/administration/organizations-teams/agent-accounts.md)
L78 in content/docs/administration/organizations-teams/agent-accounts.md "There is a Pulumi documentation page for Pulumi Cloud accounts at /docs/administration/organizations-teams/accounts/." → ✅ verified (evidence: The file content/docs/administration/organizations-teams/accounts.md exists and includes /docs/administration/organizations-teams/accounts/ in its aliases list, confirming the documentation page exists at that path.; source: repo:content/docs/administration/organizations-teams/accounts.md)
L79 in content/docs/administration/organizations-teams/agent-accounts.md "The page cross-references a blog post titled 'The agentic infrastructure era' at /blog/the-agentic-infrastructure-era/." → ✅ verified (framing: strengthened — the URL is confirmed live and the phrase "The agentic infrastructure era" is used as the blog's title/theme; exact HTML title tag not retrieved…; evidence: A Morningstar press release (May 19, 2026) confirms the URL pulumi.com/blog/the-agentic-infrastructure-era exists and is live. The phrase "The agentic infrastructure era" is used as the title/theme of this launch blog post across Pulumi'…; source: WebSearch ran query "pulumi.com/blog/the-agentic-infrastructure-era title"; Morningstar press release at https://www.morningstar.com/news/pr-newswire/20260519sf62492/pulumi-closes-the-ai-deployment-gap-with-agent-native-infrastructure-and-superintelligence-partnerships)
L80 in content/docs/administration/organizations-teams/agent-accounts.md "There is a Pulumi documentation page for direct resource operations at /docs/iac/cli/direct-resource-operations/." → ❌ contradicted (evidence: The content/docs/iac/cli/ directory on the master branch contains only: _index.md, api.md, command-line-completion.md, commands/, environment-variables.md, and exit-codes.md. There is no direct-resource-operations file or s…; source: gh api repos/pulumi/docs/contents/content/docs/iac/cli)
L3 in content/docs/iac/cli/direct-resource-operations.md "pulumi do allows creating, reading, updating, deleting, and querying resources without a project or program." → ✅ verified (evidence: The doc at line 3 (meta_desc) states: "Create, read, update, delete, and query resources without a project or program." The body confirms: "The pulumi do command provides direct operations on cloud resources through the Pulumi CLI withou…; source: content/docs/iac/cli/direct-resource-operations.md)
L14 in content/docs/iac/cli/direct-resource-operations.md "The page has an alias at /docs/iac/concepts/direct-resource-operations/." → ✅ verified (evidence: The frontmatter of the file contains an aliases block with exactly /docs/iac/concepts/direct-resource-operations/ listed.; source: content/docs/iac/cli/direct-resource-operations.md)
L18 in content/docs/iac/cli/direct-resource-operations.md "pulumi do is in research preview and the command interface may change based on feedback." → ✅ verified (evidence: The file at line 18 (inside a {{% notes type="info" %}} block) reads exactly: "pulumi do is in research preview. The command interface may change based on feedback." — a direct match to the claim.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L21 in content/docs/iac/cli/direct-resource-operations.md "The pulumi do command provides direct operations on cloud resources through the Pulumi CLI without requiring a project, program, or state file." → ✅ verified (evidence: The file at L21 reads verbatim: "The pulumi do command provides direct operations on cloud resources through the Pulumi CLI without requiring a project, program, or state file." This exactly matches the claim.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L21 in content/docs/iac/cli/direct-resource-operations.md "pulumi do exposes the full Pulumi provider ecosystem as a CLI, with commands generated dynamically from each provider's schema." → ✅ verified (evidence: The file at line 21 states verbatim: "It exposes the full Pulumi provider ecosystem as a CLI, with commands generated dynamically from each provider's schema." — this is an exact match to the claim.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L25-28 in content/docs/iac/cli/direct-resource-operations.md "pulumi do supports two types of operations: provider functions (read-only queries against cloud APIs) and resource operations (create, read, patch/update, de…" → ✅ verified (evidence: The file at lines 25-28 states: "pulumi do supports two types of operations:" with "Provider functions: Read-only queries against cloud APIs" and "Resource operations: Create, read, patch (update), delete, and list cloud resource…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L40 in content/docs/iac/cli/direct-resource-operations.md "The package, module, and type/function segments in pulumi do commands come directly from the provider schema." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md contains the exact sentence at line ~40: "The package, module, and type/function segments come directly from the provider schema." This is consistent with the broader descrip…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L40 in content/docs/iac/cli/direct-resource-operations.md "Passing --help at any level of the pulumi do command tree discovers available subcommands." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md contains the exact sentence: "Pass --help at any level of the command tree to discover available subcommands." This directly matches the claim.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L46 in content/docs/iac/cli/direct-resource-operations.md "pulumi do supports querying cloud APIs; pulumi up does not." → ✅ verified (evidence: The comparison table in the file explicitly states: "Querying cloud APIs | Yes | No" for pulumi do vs pulumi up respectively, directly confirming the claim.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L48 in content/docs/iac/cli/direct-resource-operations.md "pulumi do supports exploring a provider's capabilities; pulumi up does not." → ✅ verified (evidence: The comparison table in the file explicitly states: "Exploring a provider's capabilities | Yes | No" — confirming that pulumi do supports this capability while pulumi up does not.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L51-54 in content/docs/iac/cli/direct-resource-operations.md "pulumi do does not support state tracking and drift detection (it is stateless); pulumi up does." → ✅ verified (evidence: The file's comparison table explicitly states: "State tracking and drift detection | No (stateless) | Yes" — confirming pulumi do is stateless and does not support state tracking/drift detection, while pulumi up does.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L58 in content/docs/iac/cli/direct-resource-operations.md "Provider functions in pulumi do are read-only operations that query cloud APIs through Pulumi's provider layer." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md contains the exact sentence: "Provider functions are read-only operations that query cloud APIs through Pulumi's provider layer." — a verbatim match to the claim.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L66 in content/docs/iac/cli/direct-resource-operations.md "The input file for pulumi do provider functions contains the function's arguments, and the output is JSON written to stdout." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md (under "Running a function") states verbatim: "The input file contains the function's arguments. The output is JSON written to stdout."; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L97 in content/docs/iac/cli/direct-resource-operations.md "PCL is the default input format for pulumi do, where top-level assignments map to function parameters." → ✅ verified (evidence: The file at L97 explicitly states under "Input file formats": "PCL (default): Top-level assignments map to function parameters" and the flags table confirms --input defaults to pcl.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L106 in content/docs/iac/cli/direct-resource-operations.md "The PCL input for pulumi do is bound against the function's schema for full type checking before execution." → 🤷 unverifiable (evidence: The claim appears verbatim in the PR's own documentation file at the relevant line: "The PCL input is bound against the function's schema for full type checking before execution." However, pulumi do is described as a "research preview" f…; source: repo:content/docs/iac/cli/direct-resource-operations.md; intuition: The pulumi do command is in "research preview" — the type-checking behavior described may be aspirational/planned rat…)
L108 in content/docs/iac/cli/direct-resource-operations.md "To use YAML input with pulumi do, you pass --input yaml alongside --input-file, and the CLI converts YAML to PCL through a converter plugin before evalua…" (also L114) → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md states under "Input file formats": "YAML: Pass --input yaml alongside --input-file" and "The CLI converts YAML to PCL through a converter plugin before evaluation." —…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L118 in content/docs/iac/cli/direct-resource-operations.md "Each pulumi do resource operation uses the same provider logic as a full Pulumi program." → ➖ not-a-claim (evidence: The file content/docs/iac/cli/direct-resource-operations.md is the PR author's own documentation for the pulumi do feature (marked "research preview"). The sentence "Each operation uses the same provider logic as a full Pulumi program.…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L122-123 in content/docs/iac/cli/direct-resource-operations.md "The pulumi do create operation prompts for confirmation before creating a resource." → ✅ verified (evidence: The file at the "### Create" section (L122-123) states: "Creates a new cloud resource. Pass inputs via an input file. The CLI prompts for confirmation before creating." The flags table also lists --yes as "Auto-approve confirmation promp…; source: content/docs/iac/cli/direct-resource-operations.md)
L130-135 in content/docs/iac/cli/direct-resource-operations.md "On successful create, pulumi do outputs a JSON object with id and properties fields." → ✅ verified (evidence: The file explicitly documents under the "Create" section: "Output on success: { \"id\": \"<provider-assigned-id>\", \"properties\": { ... } }". The "Output format" section also states: "Resource operations return a JSON object with id…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L139 in content/docs/iac/cli/direct-resource-operations.md "The pulumi do read operation reads the current state of an existing resource by its cloud provider ID." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md under the "Read" section states verbatim: "Reads the current state of an existing resource by its cloud provider ID." This is an exact match to the claim.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L147 in content/docs/iac/cli/direct-resource-operations.md "The pulumi do patch (update) operation reads the current state, merges your changes, displays a diff, and prompts for confirmation." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md in the "Patch (update)" section states verbatim: "Updates an existing resource. The CLI reads the current state, merges your changes, displays a diff, and prompts for confirm…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L155 in content/docs/iac/cli/direct-resource-operations.md "The pulumi do delete operation prompts for confirmation before destroying a resource." → ✅ verified (evidence: The file at the Delete section states: "Deletes a resource. The CLI prompts for confirmation before destroying." This directly confirms the claim.; source: content/docs/iac/cli/direct-resource-operations.md)
L163 in content/docs/iac/cli/direct-resource-operations.md "The pulumi do list operation lists resources of a given type only when the provider supports listing." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md under the "List" section states: "Lists resources of a given type (when the provider supports listing)." — exactly matching the claim.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L171-172 in content/docs/iac/cli/direct-resource-operations.md "The --all flag for pulumi do list retrieves all resources with no pagination limit." → ➖ not-a-claim (evidence: The file at L171-172 documents the --all flag as "Retrieve all resources (no pagination limit)" in a table describing the PR author's own pulumi do list command, which is explicitly marked as "research preview." This is a faithful desc…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L179 in content/docs/iac/cli/direct-resource-operations.md "The --input flag for pulumi do defaults to pcl; other formats like yaml require a converter plugin." → ➖ not-a-claim (evidence: The claim is a faithful description of the PR author's own design: the flags table in the same file (line ~179) defines --input with default pcl and notes "other formats like yaml require a converter plugin." This is the author docum…; source: repo:content/docs/iac/cli/direct-resource-operations.md)

----- PINNED-COMMENT-DELIMITER -----

continued from previous comment

github-actions · 2026-05-20T20:20:10Z

continued from previous comment

L181-184 in content/docs/iac/cli/direct-resource-operations.md "The --dry-run flag for pulumi do defaults to false and runs in preview mode where the provider returns placeholder values." → ✅ verified (evidence: The flags table in the file explicitly lists: --dry-run | bool | false | Run in preview mode (provider returns placeholder values), which exactly matches the claim's description of the default value and behavior.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L181 in content/docs/iac/cli/direct-resource-operations.md "The --provider-format flag for pulumi do defaults to pcl." → ✅ verified (evidence: The flags table in the file at the relevant section explicitly states: --provider-format | string | pcl | Format of the provider configuration file, confirming the default is pcl.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L188 in content/docs/iac/cli/direct-resource-operations.md "All pulumi do operations write structured JSON to stdout, while progress messages and prompts go to stderr." → ➖ not-a-claim (framing: strengthened — the claim is a verbatim restatement of the PR author's own design documentation for a research-preview feature; no external authoritative source…; evidence: The text at L188 in content/docs/iac/cli/direct-resource-operations.md reads: "All pulumi do operations write structured JSON to stdout. Progress messages and prompts go to stderr." This is the PR author's own description of the behavi…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L188 in content/docs/iac/cli/direct-resource-operations.md "pulumi do progress messages and prompts go to stderr." → ➖ not-a-claim (framing: not-a-claim — this is a faithful description of the PR author's own design for the pulumi do command (research preview), not a third-party-attributed asserti…; evidence: The file itself (content/docs/iac/cli/direct-resource-operations.md) states "Progress messages and prompts go to stderr" as part of the Output format section for pulumi do, which is described as a "research preview" feature. This is the…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L198 in content/docs/iac/cli/direct-resource-operations.md "Secrets appear as [secret] in pulumi do output by default; use --show-secrets to reveal them." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md states verbatim: "Secrets appear as [secret] in output by default. Use --show-secrets to reveal them." The flags table also confirms --show-secrets with default false.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L200 in content/docs/iac/cli/direct-resource-operations.md "Provider functions in pulumi do return the raw function result as JSON, while resource operations return a JSON object with id and properties fields." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md contains the exact sentence in the "Output format" section: "Provider functions return the raw function result as JSON. Resource operations return a JSON object with id and…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L200 in content/docs/iac/cli/direct-resource-operations.md "pulumi do resource operations return a JSON object with id and properties fields." → ✅ verified (evidence: The file explicitly states: "Resource operations return a JSON object with id and properties fields." This is also reinforced by the Create operation's documented output: {"id": "<provider-assigned-id>", "properties": { ... }}.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L204-207 in content/docs/iac/cli/direct-resource-operations.md "pulumi do resolves provider configuration through ambient credentials (environment variables and credential files already present in the shell, e.g., AWS_AC…" → ✅ verified (evidence: The doc at L204-207 states: "1. **Ambient credentials**: Environment variables and credential files already present in the shell (e.g., AWS_ACCESS_KEY_ID, ~/.aws/credentials`). 1. Provider configuration file: Supply provider config…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L220-224 in content/docs/iac/cli/direct-resource-operations.md "Planned stateful mode for pulumi do will enable drift detection on resources created with pulumi do, lifecycle management (refresh, protect, retain-on-dele…" → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md under "## Stateful mode (planned)" states: "Stateful mode will persist resource state across operations, enabling: - Drift detection on resources created with pulumi do - L…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L226 in content/docs/iac/cli/direct-resource-operations.md "In stateful mode (planned), each mutation stores a PCL fragment in the stack's snapshot, and the collection of fragments forms an embedded program that the eng…" → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md contains the exact text: "In stateful mode, each mutation stores a PCL fragment in the stack's snapshot. The collection of fragments forms an embedded program that the engine…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L228 in content/docs/iac/cli/direct-resource-operations.md "In stateful mode (planned), on first mutation, pulumi do automatically creates a default project at $PULUMI_HOME/do/default/ with a dev stack." → ✅ verified (evidence: The file content/docs/iac/cli/direct-resource-operations.md contains the exact text at the relevant line: "On first mutation, pulumi do automatically creates a default project at $PULUMI_HOME/do/default/ with a dev stack." — within…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L232 in content/docs/iac/cli/direct-resource-operations.md "pulumi do serves as an on-ramp to the full Pulumi IaC workflow. The planned progression:" → ✅ verified (evidence: The claim text appears verbatim in the file at the stated line: "pulumi do serves as an on-ramp to the full Pulumi IaC workflow." The linked path /docs/iac/ resolves to content/docs/iac/_index.md, which exists and is ti…; source: repo:content/docs/iac/cli/direct-resource-operations.md and repo:content/docs/iac/_index.md)
L235 in content/docs/iac/cli/direct-resource-operations.md "1. Accumulate state: Stateful mode tracks resources and their relationships." → ➖ not-a-claim (framing: The regex trigger on "Accumulate" flagged this as an attribution claim, but the line is a faithful description of the PR author's own design for a new CLI feat…; evidence: No external source uses the specific phrase "Accumulate state" as a named step in a Pulumi "stateful mode." The line appears to be the PR author's own description of a new direct-resource-operations feature being documented, not an attribu…; source: WebSearch ran query "Pulumi stateful mode accumulate state tracks resources relationships" and "Pulumi direct resource operations stateful mode accumulate state"; top results didn't address the claim.)
L236 in content/docs/iac/cli/direct-resource-operations.md "The planned pulumi init --import-from-do command generates a Pulumi project with resources imported and dependency-tracked." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md explicitly states under "Graduation to infrastructure as code": "The planned progression: ... Eject to a full project: Run pulumi init --import-from-do to generate a Pu…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L239 in content/docs/iac/cli/direct-resource-operations.md "pulumi do uses the same providers and resource types as pulumi up." → ✅ verified (evidence: The file at content/docs/iac/cli/direct-resource-operations.md contains the exact statement: "pulumi do uses the same providers and resource types as pulumi up, so there's nothing to migrate."; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L245 in content/docs/iac/cli/direct-resource-operations.md "The --env flag for Pulumi ESC credential resolution is not yet available in pulumi do." → ✅ verified (evidence: The file at line ~245 contains the comment , directly confirming that the --env flag for ESC credential resolution is not yet available in pulumi do.; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L246 in content/docs/iac/cli/direct-resource-operations.md "- No state tracking: Stateless mode does not record what resources you create." → ➖ not-a-claim (evidence: The statement is the PR author's own description of a feature ("stateless mode") being introduced in their documentation file. It faithfully describes the design intent: stateless mode bypasses Pulumi's normal state recording. No external…; source: WebSearch ran query "Pulumi direct resource operations stateless mode CLI docs"; no external source contradicts or attributes this claim — it is the author's own design description.)
L247 in content/docs/iac/cli/direct-resource-operations.md "- No inline flags for properties: You must provide an input file. CLI flag-based property input is not yet supported." → ✅ verified (evidence: The direct-resource-operations.md file consistently requires --input-file for all resource and function inputs (create, patch, provider functions), with no inline property flag mechanism described anywhere. The flags table lists only `…; source: repo:content/docs/iac/cli/direct-resource-operations.md)
L249 in content/docs/iac/cli/direct-resource-operations.md "pulumi do does not appear in pulumi --help output; it must be accessed directly by name." → 🤷 unverifiable (evidence: verification did not converge within 8 turns)
L253-255 in content/docs/iac/cli/direct-resource-operations.md "The resource management with Pulumi IaC documentation is located at /docs/iac/concepts/resources/." → ✅ verified (evidence: The file content/docs/iac/concepts/resources/_index.md exists in the repository with title "Resources" and meta description about Pulumi resources, confirming the documentation path /docs/iac/concepts/resources/ is valid.; source: repo:content/docs/iac/concepts/resources/_index.md)
L256 in content/docs/iac/cli/direct-resource-operations.md "There is a blog post titled 'Introducing pulumi do' at /blog/pulumi-do-direct-resource-operations/." → ✅ verified (evidence: The file content/blog/pulumi-do-direct-resource-operations/index.md exists with title: "Introducing pulumi do: Direct Resource Operations for Any Cloud", confirming the blog post at the claimed path with a title matching "Introducing `…; source: repo:content/blog/pulumi-do-direct-resource-operations/index.md)

📊 Editorial balance

Single-subject post; balance check N/A.

🚨 Outstanding in this PR

No outstanding findings — all issues resolved.

⚠️ Low-confidence

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

[L5] content/blog/pulumi-do-direct-resource-operations/index.md — meta description describes pulumi do capabilities. Web search couldn't find external coverage because the command is being launched in this PR; the in-repo docs page (content/docs/iac/cli/direct-resource-operations.md, added in this same PR) is consistent. Author question: once the CLI release lands publicly, please confirm the description still matches the shipped command surface.
[L41] content/blog/pulumi-do-direct-resource-operations/index.md — the "smallest possible commitment / progressively level up" framing attributed to Joe Duffy's "agentic infrastructure era" post couldn't be located via web search (the cited post exists but the specific principle wasn't surfaced). Author question: is this paraphrased? If so, consider rewording as "Joe Duffy outlined" instead of "framed the key design principle" so the attribution is softer.
[L41] content/blog/pulumi-do-direct-resource-operations/index.md — the "redesigned CLI so both humans and agents can find the right command on the first try" framing reads as stronger than what the cited "Better CLI Interactions for Agents and Humans" post probably says. Web search confirmed the post exists but couldn't surface that specific framing. Author question: does the cited post actually use language like "redesigned" or "right command on the first try"? If not, soften to e.g., "improved the CLI's agent ergonomics" with a link to the same post.
[L53] content/blog/pulumi-do-direct-resource-operations/index.md — narrative claim about where AI agents stall (creating cloud accounts, plumbing credentials, wiring configuration). This is an authorial framing rather than a citable fact; leaving as-is is fine. Author question: consider grounding in a concrete example to make the claim concrete (e.g., a specific tool or scenario where this friction shows up).
[L79] content/blog/pulumi-do-direct-resource-operations/index.md — claim that "Provider functions shipped in Pulumi CLI v3.242.0". The GitHub release page returned the SPA shell during verification, so the release notes couldn't be confirmed. Author question: confirm the version number against the release notes before publish; if v3.242.0 is correct, no fix is needed.
[L168] content/blog/pulumi-do-direct-resource-operations/index.md — claim about --all and --count N semantics for pulumi do list. The companion docs page (direct-resource-operations.md L171-172) documents the same flags, so internal consistency is good. Author question: verify against the shipped CLI that the precedence between --all and --count N is documented as intended.
[L185] content/blog/pulumi-do-direct-resource-operations/index.md — zero-to-cloud-resources-in-one-session story. The doc page (agent-accounts.md) does describe this flow, but "no CLI installed" is a separate assumption (the CLI install step still has to happen). Author question: does the agent account flow include CLI bootstrap, or is the reader expected to have the CLI already? Worth clarifying so the "from zero" claim is unambiguous.
[L189] content/blog/pulumi-do-direct-resource-operations/index.md — claim that the <package> <module> <type> <operation> pattern is consistent across AWS, Azure, GCP, Kubernetes, Cloudflare, Datadog, etc. The cited better-cli-interactions-for-agents-and-humans post wasn't surfaced by web search. Author question: is that blog post published yet? If it's still in flight, link to a stable reference (e.g., the docs page added in this PR) instead.
[L211] content/blog/pulumi-do-direct-resource-operations/index.md — planned --env flag and ESC behaviors. The companion docs page has a  marker (direct-resource-operations.md L213), so this is forward-looking. Author question: consider explicitly marking this as a roadmap item in the blog post (the framing "We're building…" helps but the OIDC / deep-merge detail reads as concrete).
[L236] content/blog/pulumi-do-direct-resource-operations/index.md — claim that pulumi convert from pulumi do state will produce subnet.vpcId = myVpc.id (an expression) rather than a hard-coded string. This is roadmap behavior. Author question: confirm with engineering that the convert path preserves expression form (not literal IDs) — this is the load-bearing detail of the section.
[L248] content/blog/pulumi-do-direct-resource-operations/index.md — planned pulumi init --import-from-do behavior. The docs page (direct-resource-operations.md L236) describes the same flow. Author question: is the --import-from-do flag name final? If subject to change, hedge slightly ("a planned pulumi init flag" rather than naming it).
[L45] content/docs/administration/organizations-teams/agent-accounts.md — claim that the claim flow transfers state, history, and resources automatically "with no migration". External sources don't yet document this. Author question: confirm the no-migration guarantee with the agent-accounts engineering owner before publish (it's a strong absolute claim).
[L47] content/docs/administration/organizations-teams/agent-accounts.md — "Claiming never reduces capability" is a strong, absolute guarantee. Author question: if there's any edge case where capability is temporarily reduced (org-membership reshuffling, billing setup, etc.), consider softening to "Claiming preserves all existing capabilities" — same intent, less absolute.
[L53] content/docs/administration/organizations-teams/agent-accounts.md — the 30-day claim window number. External coverage (The New Stack) describes a 72-hour expiry but doesn't address the longer claim window. The doc itself is internally consistent: 72-hour active period → read-only → 30-day claim window. Author question: confirm the 30-day claim window with the agent-accounts owner; if it differs (e.g., 14 days, 90 days), update L30, L53, and L73 together.
[L55] content/docs/administration/organizations-teams/agent-accounts.md — strong safety guarantee that cloud resources are never touched if the Pulumi Cloud account permanently locks. This is technically true because Pulumi Cloud doesn't have credentials in users' cloud accounts, but the wording reads as a promise. Author question: consider clarifying why (e.g., "Pulumi Cloud never holds your cloud-provider credentials, so resources continue running in your account regardless of the Pulumi Cloud state") so readers see the mechanism, not just the guarantee.
[L73] content/docs/administration/organizations-teams/agent-accounts.md — read-only intermediate state and 30-day restore window. External sources confirm 72-hour expiry but not the read-only mode or 30-day claim period. Author question: confirm the read-only state (vs. fully locked) with the agent-accounts owner — and the 30-day window, which should match L30 and L53.
[L106] content/docs/iac/cli/direct-resource-operations.md — claim that PCL is bound against the function schema for full type checking. The blog post (L63) makes a similar claim ("same type checking and default handling that a full Pulumi program provides"), so the two are consistent. Author question: since pulumi do is research preview, confirm "full type checking" reflects what's currently shipped (vs. planned).
[L249] content/docs/iac/cli/direct-resource-operations.md — claim that pulumi do is hidden from pulumi --help. This is a research-preview UX choice; worth flagging so readers don't assume the command is missing. Author question: if the command becomes generally available, this limitation should be removed — track that as a follow-up.

Style findings

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

_{Click each filename to expand.}

content/blog/pulumi-do-direct-resource-operations/index.md (2 issues: 1 wordiness, 1 difficulty qualifier)

line 51: [style] wordiness — 'it is' is too wordy.
line 53: [style] difficulty qualifier — Avoid difficulty qualifier 'just' -- it judges difficulty for the reader (STYLE-GUIDE.md §Inclusive Language).

content/docs/administration/organizations-teams/agent-accounts.md (3 issues: 2 wordiness, 1 punctuation)

line 18: [style] wordiness — 'it is' is too wordy.
line 67: [style] punctuation — Use the Oxford comma in 'If multiple agents run on the same machine (for example, Cursor and'.
line 71: [style] wordiness — 'it is' is too wordy.

content/docs/iac/cli/direct-resource-operations.md (1 issues: 1 substitution)

line 188: [style] substitution — Use 'navigate to' instead of 'go to' (STYLE-GUIDE.md).

📋 Triaged verifier findings

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

[L30] content/docs/administration/organizations-teams/agent-accounts.md — "The agent account claim window is 30 days from creation." Spurious: the verifier compared this against external coverage of the 72-hour active period, but the doc explicitly distinguishes a 72-hour active period (L51) from a 30-day claim window (L53). The two figures describe different things, so the 72-hour figure does not contradict the 30-day claim window. (The 30-day window is still unverified externally; see the ⚠️ L53 entry.)
[L30] content/docs/administration/organizations-teams/agent-accounts.md — "The agent account claim window is 30 days." Spurious: duplicate of the preceding L30 entry — same content, same triage.
[L51-53] content/blog/pulumi-do-direct-resource-operations/index.md — "Andrej Karpathy wrote: 'I didn't even spend all that much work in the code editor itself…'" Spurious: duplicate of the L51-53 entry already kept under 🚨 Outstanding — same claim, same fix.
[L80] content/docs/administration/organizations-teams/agent-accounts.md — "There is a Pulumi documentation page for direct resource operations at /docs/iac/cli/direct-resource-operations/." Spurious: the verifier checked the master branch and didn't find the file. This PR adds content/docs/iac/cli/direct-resource-operations.md in the same commit — the link target is valid against the PR head.
[L31] content/docs/administration/organizations-teams/agent-accounts.md — "https://app.pulumi.com/claim/abc123" Mis-sourced: this URL appears inside a fenced CLI-output example demonstrating what the claim-link prompt looks like. It's not a verifiable assertion — the abc123 token is a placeholder, exactly as the verifier inferred.
[L261] content/blog/pulumi-do-direct-resource-operations/index.md — "The documentation for pulumi do is at /docs/iac/cli/direct-resource-operations/." Mis-sourced: the verifier followed the GitHub release URL (truncated SPA), not the in-PR docs page. The docs page is added by this same PR (see ✅ L256 verdict in the trail).
[L265] content/blog/pulumi-do-direct-resource-operations/index.md — "- Documentation" Mis-sourced: verification ran out of turns on a path the PR creates in this same commit. The docs page exists in the PR head; the link resolves once merged.

💡 Pre-existing issues in touched files (optional)

No pre-existing issues in touched files.

✅ Resolved since last review

[L51-53] content/blog/pulumi-do-direct-resource-operations/index.md — Karpathy quote removed entirely; the section was rewritten without the attribution. (resolved in 60f9989)
[L193] content/blog/pulumi-do-direct-resource-operations/index.md — "Datadog's CLI is CI-only." rewritten to "Datadog's CLI focuses on CI/CD and serverless instrumentation rather than general resource management." (resolved in 60f9989)
[L3] content/docs/administration/organizations-teams/agent-accounts.md — "deployments" removed from meta description; now accurately reads "state and ESC without manual signup." (resolved in 60f9989)
[L57] content/blog/pulumi-do-direct-resource-operations/index.md — Eight-steps enumeration completed with eight distinct steps: create a directory, initialize a project, pick a language, install packages, configure credentials, write infrastructure code, preview, deploy. (resolved in 60f9989)

📜 Review history

2026-05-20T20:09:31Z — Initial review: 4 outstanding factual issues (Karpathy quote not found in cited source; Datadog CI-only overclaim; agent-accounts meta_desc mentions deployments not listed in body; "eight steps" enumerates six); 18 low-confidence claims about preview/planned features pending engineering confirmation. (05c5353)
2026-05-20T22:30:22Z — re-reviewed after fix push (1 new commit, 60f9989): all 4 outstanding issues resolved; no new findings introduced. (@arunkumar611)

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.)

…s meta_desc - Enumerate all eight IaC steps (was listing six) - Replace inaccurate "Datadog's CLI is CI-only" with accurate description - Remove "deployments" from agent accounts meta_desc (body only lists state and ESC) - Clean up Karpathy quote section (already addressed in prior edit) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

arunkumar611 · 2026-05-20T22:22:48Z

@claude #update-review

github-actions · 2026-05-20T22:37:42Z

🤖 Review updated on @arunkumar611's request.

… link Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

cnunciato

…rations

arunkumar611 temporarily deployed to testing May 20, 2026 20:05 — with GitHub Actions Inactive

arunkumar611 temporarily deployed to production May 20, 2026 20:05 — with GitHub Actions Inactive

arunkumar611 requested review from Frassle, cnunciato and pgavlin May 20, 2026 20:07

cnunciato added the needs-design Needs input from design/UX label May 20, 2026

arunkumar611 requested a review from CamSoper May 20, 2026 20:09

github-actions Bot added review:outstanding-issues Claude review completed; outstanding has author-actionable findings and removed review:in-progress Claude review is currently running labels May 20, 2026

arunkumar611 temporarily deployed to testing May 20, 2026 21:55 — with GitHub Actions Inactive

arunkumar611 temporarily deployed to production May 20, 2026 21:55 — with GitHub Actions Inactive

github-actions Bot added review:stale New commits since last Claude review; refresh on next ready-transition or @claude mention and removed review:outstanding-issues Claude review completed; outstanding has author-actionable findings labels May 20, 2026

github-actions Bot added review:in-progress Claude review is currently running and removed review:stale New commits since last Claude review; refresh on next ready-transition or @claude mention labels May 20, 2026

github-actions Bot removed the review:in-progress Claude review is currently running label May 20, 2026

jkeiser had a problem deploying to testing May 22, 2026 03:26 — with GitHub Actions Failure

jkeiser temporarily deployed to production May 22, 2026 03:26 — with GitHub Actions Inactive

arunkumar611 requested review from Frassle and cnunciato May 22, 2026 04:05

lint fix

c70438f

jkeiser temporarily deployed to testing May 22, 2026 04:17 — with GitHub Actions Inactive

jkeiser temporarily deployed to production May 22, 2026 04:17 — with GitHub Actions Inactive

Refine agent accounts page: tighten flow, update limitations, add Neo…

6af3659

… link Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

arunkumar611 temporarily deployed to testing May 22, 2026 22:16 — with GitHub Actions Inactive

arunkumar611 temporarily deployed to production May 22, 2026 22:16 — with GitHub Actions Inactive

Chris's edits

89426a7

cnunciato temporarily deployed to production May 23, 2026 01:14 — with GitHub Actions Inactive

cnunciato temporarily deployed to testing May 23, 2026 01:14 — with GitHub Actions Inactive

cnunciato approved these changes May 23, 2026

View reviewed changes

cnunciato added 2 commits May 22, 2026 18:28

Merge remote-tracking branch 'origin/master' into pulumi-resource-ope…

8cc3292

…rations

Links

5e2168e

cnunciato temporarily deployed to testing May 23, 2026 01:32 — with GitHub Actions Inactive

cnunciato had a problem deploying to production May 23, 2026 01:32 — with GitHub Actions Error

Links

3427e4d

cnunciato temporarily deployed to testing May 23, 2026 01:37 — with GitHub Actions Inactive

cnunciato had a problem deploying to production May 23, 2026 01:37 — with GitHub Actions Error

cnunciato enabled auto-merge (squash) May 23, 2026 01:37

cnunciato disabled auto-merge May 23, 2026 01:41

Address review nits

d435ed1

cnunciato temporarily deployed to testing May 23, 2026 01:43 — with GitHub Actions Inactive

cnunciato temporarily deployed to production May 23, 2026 01:43 — with GitHub Actions Inactive

cnunciato enabled auto-merge (squash) May 23, 2026 01:43

cnunciato merged commit 33be23e into master May 23, 2026
10 checks passed

cnunciato deleted the pulumi-resource-operations branch May 23, 2026 01:54

cnunciato temporarily deployed to testing May 23, 2026 01:54 — with GitHub Actions Inactive

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add pulumi do blog post, docs page, and agent accounts#19291

Add pulumi do blog post, docs page, and agent accounts#19291
cnunciato merged 15 commits into
masterfrom
pulumi-resource-operations

arunkumar611 commented May 20, 2026 •

edited by cnunciato

Loading

Uh oh!

github-actions Bot commented May 20, 2026 •

edited

Loading

Uh oh!

pulumi-bot commented May 20, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented May 20, 2026 •

edited by pulumi-bot

Loading

Uh oh!

github-actions Bot commented May 20, 2026 •

edited by pulumi-bot

Loading

Uh oh!

arunkumar611 commented May 20, 2026

Uh oh!

github-actions Bot commented May 20, 2026

Uh oh!

cnunciato left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Conversation

arunkumar611 commented May 20, 2026 • edited by cnunciato Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Uh oh!

github-actions Bot commented May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Social Media Review

content/blog/pulumi-do-direct-resource-operations/index.md

X — missing

LinkedIn — missing

Bluesky — missing

Suggested copy

Uh oh!

pulumi-bot commented May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented May 20, 2026 • edited by pulumi-bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pre-merge Review — Last updated 2026-05-20T22:30:22Z

🔍 Verification trail

Uh oh!

github-actions Bot commented May 20, 2026 • edited by pulumi-bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

📊 Editorial balance

🚨 Outstanding in this PR

⚠️ Low-confidence

Style findings

📋 Triaged verifier findings

💡 Pre-existing issues in touched files (optional)

✅ Resolved since last review

📜 Review history

Uh oh!

arunkumar611 commented May 20, 2026

Uh oh!

github-actions Bot commented May 20, 2026

Uh oh!

cnunciato left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

arunkumar611 commented May 20, 2026 •

edited by cnunciato

Loading

github-actions Bot commented May 20, 2026 •

edited

Loading

pulumi-bot commented May 20, 2026 •

edited

Loading

github-actions Bot commented May 20, 2026 •

edited by pulumi-bot

Loading

github-actions Bot commented May 20, 2026 •

edited by pulumi-bot

Loading