AIToday
Large Language ModelsAI Coding AssistantsZenn AI/MLPublished: Oct 15, 2026, 22:00 JST

Claude Code 2.37337.1 update: gripes stay on X, bugs pile up on GitHub

Claude Code 2.37337.1 update: gripes stay on X, bugs pile up on GitHub

3 Key Points

  1. What happened

    After Anthropic's Claude Code desktop app update 2.37337.1, an AI-led search of X turned up user complaints about UI changes, but GitHub's anthropics/claude-code repo showed 76 issues citing the version number.

  2. Why it matters

    The AI's first headline claimed no serious bugs were found — a conclusion the body says flipped once the search moved to GitHub, suggesting where you look shapes what an AI reports.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

The piece is a first-hand record of an AI-assisted research attempt that initially misled its own user. Asked to survey bug reports and complaints about Anthropic's Claude Code desktop update, the assistant returned a headline saying no serious problems were found, and attributed the dissatisfaction to UI changes. Then the author's own screen produced an error toast, and switching to GitHub's public issue tracker for anthropics/claude-code surfaced a different picture: reports of the same toast across Linux, macOS and Windows, a migration that left most project folders lumped under 'Default', and a sidebar grouping option that disappeared.

The author's method also exposes where AI research tools stall. Web search could not effectively reach X or Reddit, the built-in browser stopped at X's login screen, Reddit and 5ch were blocked by network policy or lacked a retrieval method, and a Chrome extension returned only the first search result. The most productive move was not a search at all but reading the replies to the official announcement post, where the top-voted request asked to bring back the effort slider. Counting issues by version number, using GitHub's command-line tool, was a deliberate choice to avoid unrelated reports — and the author notes the figure is a floor, since reports that omit the number are missed.

The body also flags what remains open: reactions on Reddit and 5ch were not seen, no official comment from Anthropic had appeared on any of the eight issues cited, and the cause of the toast is only a guess by reporters, one of whom explicitly called their reading a clue rather than proof. The author's own takeaway is practical — the next time an unfamiliar toast appears, check GitHub's issues first.

FAQ
What did the AI report the first time?
It concluded that within what was visible on X, no serious bug reports were found, and complaints centered on features like UI changes. The body says that headline was wrong and its caveat — that bugs weren't showing up prominently — was what held.
What did the GitHub issues actually show?
Reports including an error toast on new sessions (#102135) seen on Linux, macOS and Windows, and a migration issue where only 8 of about 20 project folders got environments. The count of 76 issues included 72 still open.
Was there a workaround for the toast error?
One issue reported that if a session won't start, selecting 'Local' again from the environment picker after the toast can help. The author did not try it and no official fix was confirmed.

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleEiliko toy leaks 13 million conversations on Chinese servers