Test: Superpowers, GSD, and GSTACK blog post (social review test) by adamgordonbell · Pull Request #18516 · pulumi/docs

adamgordonbell · 2026-04-14T12:29:01Z

Summary

Test PR to trigger the social media review workflow on Engin's frameworks blog post
Rebased from Add blog post: AI coding agent orchestration frameworks compared #18493 onto master
Used to validate the updated social review skill

Test plan

Verify social media review workflow triggers
Check that the review output follows the updated rubric and writing guide

…compared New blog post comparing three community frameworks for AI coding agents. Covers Superpowers (TDD enforcement), GSD (context rot prevention), and GSTACK (role-based governance) with command reference tables, install instructions, and Pulumi workflow integration guidance.

Add linkedin and bluesky social keys to frontmatter. Add github-card shortcodes for the three framework repos in the getting started section.

Fix rule-of-three parallel structure, add personal opinion, vary framework descriptions, fix grammar in Twitter copy.

Address social-review feedback: replace "I tried" with neutral voice for corporate account, split Twitter into 2 paragraphs, add paragraph break to Bluesky for scannability.

Use Pulumi brand voice (no first-person), and open with the reader's pain point rather than the article's conclusions. Co-authored-by: Adam Gordon Bell <adamgordonbell@users.noreply.github.com>

Co-authored-by: Adam Gordon Bell <adamgordonbell@users.noreply.github.com>

Dan Blanchard (chardet maintainer since 2012) used Superpowers to rewrite chardet v7.0.0. Jesse Vincent created Superpowers but did not write or release chardet. Fix attribution in both blog body and LinkedIn social copy.

claude · 2026-04-14T12:32:25Z

Docs Review — PR #18516

Reviewed the new blog post content/blog/claude-code-orchestration-frameworks/index.md and the SVG asset changes. Overall this is a well-written, engaging comparison post with strong structure. A few items to address:

Issues

STYLE-GUIDE.md violation — "simple" (index.md, line 44)

"The core idea is simple: no production code gets written without a failing test first."

Per STYLE-GUIDE.md: "Avoid words like 'easy' or 'simple.' These judge difficulty and may alienate readers." Consider rephrasing:
```
The core idea is strict: no production code gets written without a failing test first.
```
First mentions of tools missing hyperlinks (index.md, line 37)

"All three started with Claude Code but now work across Cursor, Codex, Windsurf, Gemini CLI, and more."

The blog review criteria require that the first mention of every tool or technology be hyperlinked. Claude Code, Cursor, Codex, Windsurf, and Gemini CLI are all mentioned here without links. At minimum, link Claude Code to its product page and the others to their respective sites.
First mention of Y Combinator not linked (index.md, line 97)

"created by Garry Tan (CEO of Y Combinator)"

"Y Combinator" should be hyperlinked on first mention.
First mention of Playwright not linked (index.md, line 119)

"Real Playwright browser testing, not simulated"

"Playwright" should link to its project page.
GitHub star counts are volatile claims (index.md, lines 39, 63, 97, and comparison table)
The post cites specific star counts (149K, 51K, 71K) as facts. These numbers change daily. Consider either:
- Removing exact counts and using relative descriptors ("widely adopted," "popular"), or
- Adding a "as of April 2026" qualifier, or
- Accepting they will be stale quickly (common in blog posts, but worth a conscious decision)
Internal link /docs/iac/clouds/aws/guides/vpc/ uses trailing slash (index.md, line 56)
The actual file is content/docs/iac/clouds/aws/guides/vpc.md, which Hugo will serve at /docs/iac/clouds/aws/guides/vpc/. This should resolve correctly, but verify the link works in a preview since it is a leaf file, not a section with _index.md.

Same applies to /docs/iac/clouds/aws/guides/iam/ (line 137) — the file is iam.md, not iam/_index.md.

Minor Observations

Title length: 79 characters, but allow_long_title: true is correctly set. ✅
Meta description: ~140 characters, well under the 160 limit. ✅
 break: Present after the intro paragraphs. ✅
Author profile: engin-diri.toml exists with avatar at static/images/team/engin-diri.jpg. ✅
social: block: Present with twitter, linkedin, and bluesky copy. ✅
meta_image: Set to meta.png (320KB), displays correctly with Pulumi branding. ✅
Heading case: H1 in Title Case, H2+ in sentence case. ✅
Code blocks: Bash block has language specifier. ✅
Shortcodes: github-card and blog/cta-button both exist in the site. ✅
Image sizes: meta.png (320KB) and feature.png (154KB) are reasonable. ✅
Writing quality: Strong, opinionated voice. Good sentence variety. No excessive hedging or AI writing patterns. The comparison tables are well-structured and add real value.

SVG Asset Changes

The PR replaces the verbose OpenAI SVG with a compact single-path version, and adds new Anthropic and Cursor SVGs. These are internal tooling assets (.claude/commands/blog-meta-image/assets/logos/) and look correct.

Publishing Readiness Checklist

Mention me (@claude) if you'd like additional reviews or fixes after addressing these items.

pulumi-bot · 2026-04-14T12:34:53Z

Your site preview for commit 09e3648 is ready! 🎉

http://www-testing-pulumi-docs-origin-pr-18516-09e36484.s3-website.us-west-2.amazonaws.com

Two bugs prevented the social review from running: 1. bash -e killed the check step when the schedule-posts script exited non-zero (e.g., character limit exceeded), so EXIT=$? never ran and the Claude review step was skipped entirely. 2. github_token was set as an env var but not passed as an action input to claude-code-action, which needs it to function.

github-actions · 2026-04-14T12:41:19Z

Social Media Review

content/blog/claude-code-orchestration-frameworks/index.md

X — FAIL

Over character limit: 386-char body exceeds the 255-char limit
Summarizes the article: names all three failure modes (context rot, skipped tests, scope drift) and gives away the selection criterion
Weak opening: "AI coding agents fail in recognizable ways" is generic

LinkedIn — FAIL

Weak opening: "There is a pattern that teams tend to hit" is generic filler
Curiosity gap closed: "which helps most comes down to which problem keeps showing up for you" gives away the article's framing
Lacks specificity: no concrete numbers, names, or results anywhere in the copy

Bluesky — FAIL

Over character limit: 421-char body exceeds the 300-char limit
Summarizes the article: same three failure modes listed, same selection criterion given
Weak opening: "AI coding agents tend to work well at first" is generic

Suggested copy

X (222/255 chars):

We asked a coding agent for a three-subnet VPC. It added a NAT gateway, transit gateway, VPN endpoint, and DNS resolver on its own.

Three frameworks claim to fix this. One produced a 41x speedup. Here's how they compared.

LinkedIn (549/2950 chars):

We asked a coding agent for a three-subnet VPC. It added a NAT gateway, transit gateway, VPN endpoint, and a DNS resolver. Infrastructure nobody asked for and we would pay for monthly.

That is scope drift. There are two more failure patterns just as predictable. Three community frameworks each fix a different one.

We spent a few weeks running all three on real Pulumi infrastructure work. One produced a 41x speedup on a real library rewrite. One we probably would not reach for on pure infrastructure work. Here is where each one actually fits.

Bluesky (277/300 chars):

We asked a coding agent for a three-subnet VPC. It added a NAT gateway, transit gateway, VPN endpoint, and DNS resolver nobody requested.

Three frameworks tackle the patterns that break AI coding agents on longer sessions. One produced a 41x speedup. Here's how they compared.

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

adamgordonbell · 2026-04-14T12:43:05Z

@claude please update the social posts in the frontmatter with the suggested copy from the social media review above

pulumi-bot · 2026-04-14T12:43:46Z

Claude finished @adamgordonbell's task in 54s —— View job

Updating social media copy

Understand the request
Read the blog post frontmatter
Replace social copy with suggested copy from the social media review
Commit and push

All three social posts have been updated with the suggested copy from the social media review:

X/Twitter: Now leads with the concrete VPC scope-drift example and the 41x speedup hook (222 chars, within 255 limit)
LinkedIn: Three-paragraph structure — scope drift example, the problem pattern, and what readers will find (549 chars, within 2950 limit)
Bluesky: Tight version of the same hook (277 chars, within 300 limit)
·

Co-authored-by: Adam Gordon Bell <adamgordonbell@users.noreply.github.com>

After Claude updates social copy, the push doesn't trigger a re-review (GITHUB_TOKEN pushes don't trigger workflows). Adding issue_comment trigger lets anyone comment /social-review to re-run the review.

claude · 2026-04-14T12:53:11Z

Social Media Review

content/blog/claude-code-orchestration-frameworks/index.md

X — PASS

LinkedIn — PASS

Bluesky — PASS

dirien and others added 7 commits April 14, 2026 08:28

Add social media copy and GitHub card shortcodes

7ddefbd

Add linkedin and bluesky social keys to frontmatter. Add github-card shortcodes for the three framework repos in the getting started section.

Rewrite social media copy with author voice and humanizer fixes

529aa91

Fix rule-of-three parallel structure, add personal opinion, vary framework descriptions, fix grammar in Twitter copy.

Fix social media copy: drop first-person, split into paragraphs

35e5644

Address social-review feedback: replace "I tried" with neutral voice for corporate account, split Twitter into 2 paragraphs, add paragraph break to Bluesky for scannability.

Rewrite social posts as hooks instead of article summaries

6d0dab6

Use Pulumi brand voice (no first-person), and open with the reader's pain point rather than the article's conclusions. Co-authored-by: Adam Gordon Bell <adamgordonbell@users.noreply.github.com>

Rewrite social posts: natural language, no em dashes, complete sentences

57fffe4

Co-authored-by: Adam Gordon Bell <adamgordonbell@users.noreply.github.com>

adamgordonbell had a problem deploying to production April 14, 2026 12:29 — with GitHub Actions Failure

adamgordonbell temporarily deployed to testing April 14, 2026 12:29 — with GitHub Actions Inactive

adamgordonbell had a problem deploying to production April 14, 2026 12:35 — with GitHub Actions Failure

adamgordonbell temporarily deployed to testing April 14, 2026 12:35 — with GitHub Actions Inactive

github-actions Bot and others added 2 commits April 14, 2026 12:44

Update social media copy with suggested copy from social review

a60f0f7

Co-authored-by: Adam Gordon Bell <adamgordonbell@users.noreply.github.com>

Add /social-review comment trigger to social review workflow

09e3648

After Claude updates social copy, the push doesn't trigger a re-review (GITHUB_TOKEN pushes don't trigger workflows). Adding issue_comment trigger lets anyone comment /social-review to re-run the review.

adamgordonbell temporarily deployed to testing April 14, 2026 12:48 — with GitHub Actions Inactive

adamgordonbell had a problem deploying to production April 14, 2026 12:48 — with GitHub Actions Error

adamgordonbell mentioned this pull request Apr 14, 2026

Fix social review workflow: bash -e, github_token, /social-review trigger #18517

Merged

3 tasks

adamgordonbell closed this Apr 14, 2026

adamgordonbell deleted the test/claude-code-frameworks branch April 14, 2026 13:04

adamgordonbell mentioned this pull request Apr 14, 2026

Fix social review concurrency: prevent comments from cancelling runs #18519

Merged

4 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Test: Superpowers, GSD, and GSTACK blog post (social review test)#18516

Test: Superpowers, GSD, and GSTACK blog post (social review test)#18516
adamgordonbell wants to merge 10 commits intomasterfrom
test/claude-code-frameworks

adamgordonbell commented Apr 14, 2026

Uh oh!

claude Bot commented Apr 14, 2026

Uh oh!

pulumi-bot commented Apr 14, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Apr 14, 2026

Uh oh!

adamgordonbell commented Apr 14, 2026

Uh oh!

pulumi-bot commented Apr 14, 2026 •

edited

Loading

Uh oh!

claude Bot commented Apr 14, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

adamgordonbell commented Apr 14, 2026

Summary

Test plan

Uh oh!

claude Bot commented Apr 14, 2026

Docs Review — PR #18516

Issues

Minor Observations

SVG Asset Changes

Publishing Readiness Checklist

Uh oh!

pulumi-bot commented Apr 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Apr 14, 2026

Social Media Review

content/blog/claude-code-orchestration-frameworks/index.md

X — FAIL

LinkedIn — FAIL

Bluesky — FAIL

Suggested copy

Uh oh!

adamgordonbell commented Apr 14, 2026

Uh oh!

pulumi-bot commented Apr 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Updating social media copy

Uh oh!

claude Bot commented Apr 14, 2026

Social Media Review

content/blog/claude-code-orchestration-frameworks/index.md

X — PASS

LinkedIn — PASS

Bluesky — PASS

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

pulumi-bot commented Apr 14, 2026 •

edited

Loading

pulumi-bot commented Apr 14, 2026 •

edited

Loading