Claude Code team says Claude writes about 90% of new code
11/17/2026 — 11/18, 00:19·2 sources (all)·2 reports (all)
Key takeaways
- According to Addy Osmani on the Anthropic blog, the Claude Code team relies on Claude to write and review roughly 90% of new code, making ve…
Background & analysis
On November 17, 2026, at 18:20, an article examining how the Claude Code team approaches verification surfaced on the Hacker News front page. The post earned 105 points and drew 112 comments, pointing to sustained community interest in the topic. The original article is published at https://claude.dev/blog/how-the-claude-code-team-thinks-about-verification/ , while the discussion thread is available at https://news.ycombinator.com/item?id=50529612 .
The piece focuses on the step that follows AI-generated code: how the Claude Code team confirms that the produced result is actually correct. It lays out the team's thinking on verification, framing it as the stage in the coding workflow where the output of an AI model is checked for correctness before it can be treated as reliable.
At this point, only this single report is available, and the discussion remains anchored to the article's appearance on the Hacker News front page and the views it presents. There is no further material indicating that the team has released additional detail, new guidance, or follow-up statements on its verification approach, so the situation currently stands at the stage of a published summary of the team's perspective rather than any later development.
AI-generated from 1 reports · updated 6 hours ago
Latest turnAccording to Addy Osmani on Anthropic's official blog, Claude writes and reviews about 90% of new code on the Claude Code team, shifting the key question to how AI verification works. The article examines verification loops that give the agent concrete evidence of whether its work is correct, rather than relying on line-by-line human review before merging.
Related AI tools
- Heat index
- 157
- All sources
- 2
- All reports
- 2
- First seen
- 11 hours ago
Reports on this story headlines open the original
According to Addy Osmani on Anthropic's official blog, Claude writes and reviews about 90% of new code on the Claude Code team, shifting the key question to how AI verification works. The article examines verification loops that give the agent concrete evidence of whether its work is correct, rather than relying on line-by-line human review before merging.
DEV Community · AIAI score 59A post on how the Claude Code team approaches verification climbed Hacker News with 105 points and 112 comments. It focuses on how the team checks that agent-written code is actually correct, drawing developers into a discussion of review and testing workflows for coding agents.
Hacker News · 热榜AI score 53
Other stories people are talking about
- Heat index 74Developer Uses T3 Code to Remotely Control Claude Code via Mobile1 sources
- Heat index 72Developer Shares Workflow for Dual-Agent Development with Claude Code and Codex1 sources
- Heat index 132Claude Code Bypass Lets Agents Read Blocked .env Files2 sources
- Heat index 64Configuring Claude Code for Unsupervised Overnight Long Tasks1 sources
- Heat index 78Developer shares automated YouTube workflow built with Claude Code1 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

