I've been using Artifacts in Claude Code for everything: visual explanations of tricky code, system diagrams, quick previews of a few animation options, data analyses and dashboards I share with the team. They are a game changer for how I work with Claude. Can't wait to hear what you think!
86-account seed digest: Claude Code Artifacts, Jellyfish, and Codex Record & Replay — June 19
A June 19 Beijing-time digest from the 86 public seed accounts currently available for checking, not the full @hwwaanng following list. The scan checked 1,279 returned posts and found 19 qualifying non-retweet posts with 100+ likes, led by Boris Cherny on Claude Code Artifacts, QT9277 on Jellyfish for AI short dramas, and dotey on Codex Record & Replay.
Coverage note: this issue covers the 86 public seed accounts currently available for checking, not the full @hwwaanng following list. The scan window is June 19, 2026, 00:00-24:00 Beijing time; it found 19 non-retweet posts with 100+ likes from 9 authors.
What led the day
The strongest cluster was developer-agent work becoming easier to show, replay, and share. Boris Cherny, who works on Claude Code at Anthropic, got the biggest post of the day with 4,487 likes: he said he has been using Claude Code Artifacts for visual explanations, system diagrams, animation previews, data analyses, and team dashboards 1.
콘텐츠 카드를 불러오는 중…
The second-biggest item came from QT9277, a verified Chinese-language account that often tracks crypto, automation, and creator tooling. The post pitched Jellyfish as an open-source AI short-drama workflow: script in, storyboard and characters out, then export; the claim that stood out was a zero-basics creator reaching 4.6 million followers on Hongguo with 104 videos and a 53 million-view breakout 2.
콘텐츠 카드를 불러오는 중…
Dotey, a verified AI engineer and AI/software-engineering explainer, gave the most useful operator note of the day: OpenAI Codex now has Record & Replay on Mac, where a user performs a repeatable workflow once and Codex turns it into a reusable Skill. The example set was practical rather than abstract: expense forms, video-publishing steps, issue creation, and other UI-heavy workflows that are painful to describe in prose 3.
콘텐츠 카드를 불러오는 중…
Agent-workflow notes
Peter Steinberger, a verified developer working around OpenClaw and OpenAI, compressed the current tool debate into one line: everything is now either a fast API or a slow API. It landed because the rest of the day was full of tools that turn slow human workflows into repeatable agent actions 4.
콘텐츠 카드를 불러오는 중…
Dotey also covered baoyu-design, a Skill update that can call an AI image-generation Skill while producing slides, animated videos, or websites. The practical hook is exporting the result as PPTX with images included, then continuing in PowerPoint or Keynote instead of treating the generated artifact as a dead-end image 5.
콘텐츠 카드를 불러오는 중…
Dotey's second Claude Code Artifact post framed the feature as a team-collaboration fix: agent work that used to live inside one operator's terminal can now become a private, updating web page for debugging timelines, PR review, dependency audits, cost breakdowns, or UI options 6.
콘텐츠 카드를 불러오는 중…
Lakr233, whose public profile did not expose a usable bio in the detail payload, posted a compact Mac tooling note: Golden Gate now has forced-close logic for software that keeps running in the background 7.
콘텐츠 카드를 불러오는 중…
Creator-economy signal: AI short dramas
QT9277 returned later with the money question behind the Jellyfish post: if AI-generated short dramas can reach tens of millions of plays on Hongguo, how much can one of those hits actually earn? The post is less a verified revenue breakdown than a clear marker that tool adoption and creator monetization are now being discussed together 8.
콘텐츠 카드를 불러오는 중…
Turingou's four-post philosophy lane
Turingou's posts are often transient, so the safest way to use them is to paraphrase the argument and keep the original card nearby. The first one put today's AI-cost debate into a broader claim: the market is approaching a threshold where many people and organizations can no longer clearly distinguish between merely usable intelligence and top-tier intelligence, while non-frontier model costs may fall quickly 9.
콘텐츠 카드를 불러오는 중…
A second post split life choices into two patterns: seeking certainty inside uncertainty, or choosing uncertain challenge inside certainty. It reads like a personal decision framework rather than a tech take, but it explains why the account's AI posts often revolve around appetite for ambiguity 10.
콘텐츠 카드를 불러오는 중…
The third one separated enjoying creation from enjoying the reward of creation. The useful distinction: when the reward disappears, many people lose the drive that previously looked like grit; in that framing, obsession itself can be a valuable trait 11.
콘텐츠 카드를 불러오는 중…
The fourth was the day's pure travel card: a cruise arrival at King's Landing, with the sea wall matching the television image closely enough to make the post clear even without extra explanation 12.
콘텐츠 카드를 불러오는 중…
Visual and object posts
Jacob Titus, a verified account tied to Tutt Street, had two architecture-adjacent posts. The higher-liked one argued for the restorative effect of walking a city's oldest streets, paired with an image link 13.
콘텐츠 카드를 불러오는 중…
His second post used a garage image to make a scale argument: trucks used to fit. The post is short, but it belongs in the same visual-design lane as the old-street card 14.
콘텐츠 카드를 불러오는 중…
Sophia, a verified history, art, beauty, and culture account, led her set with an early-20th-century view of Antwerpen Centraal's interior. It was the most-liked visual artifact post in this window at 242 likes 15.
콘텐츠 카드를 불러오는 중…
Her next object post showed Phoebe Anna Traquair's 1890 Love Cup from Scotland 16.
콘텐츠 카드를 불러오는 중…
The next artifact in the same lane was a Byzantine Saint George medallion from an icon frame 17.
콘텐츠 카드를 불러오는 중…
She closed the set with the 1895 Rose Bud Egg by the House of Fabergé and Mikhail Perkhin 18.
콘텐츠 카드를 불러오는 중…
Nyarime, a verified developer and photographer account behind Naixi Networks, supplied the day's lightest visual note: instead of making zongzi for Dragon Boat Festival, the post went with ice cream 19.
콘텐츠 카드를 불러오는 중…
Bottom line
If you only open three items, start with Boris Cherny on Claude Code Artifacts, dotey on Codex Record & Replay, and QT9277's Jellyfish short-drama workflow. That gives you the day's main pattern: AI tooling is moving from one-off prompting toward reusable skills, shareable artifacts, and content pipelines that non-specialists can operate.
참고 출처
- 1Boris Cherny on Claude Code Artifacts
- 2QT9277 on Jellyfish for AI short dramas
- 3dotey on Codex Record & Replay
- 4Peter Steinberger on fast and slow APIs
- 5dotey on baoyu-design image generation
- 6dotey on Claude Code Artifacts
- 7Lakr233 on Golden Gate background app handling
- 8QT9277 on AI short-drama monetization
- 9turingou on the intelligence-cost threshold
- 10turingou on certainty and challenge
- 11turingou on creation and reward feedback
- 12turingou on arriving at King's Landing by cruise
- 13Jacob Titus on old city streets
- 14Jacob Titus on trucks and garage scale
- 15Sophia on Antwerpen Centraal railway station
- 16Sophia on The Love Cup
- 17Sophia on a Byzantine Saint George medallion
- 18Sophia on the Fabergé Rose Bud Egg
- 19Nyarime on Dragon Boat Festival ice cream
이 콘텐츠를 둘러싼 관점이나 맥락을 계속 보강해 보세요.