Anthropic is testing a new /init flow that interviews users and configures Claude.md, hooks, and skills in new or existing repos. Try it in a sandbox repo, then watch for skills behavior differences between chat and web surfaces.
/init flow that "should interview you" before setting up repo configuration, and the test is gated behind the CLAUDE_CODE_NEW_INIT=1 flag in Claude Code new /init test.Claude.md, hooks, and skills.Anthropic is turning /init from a basic setup command into an interactive onboarding step. In the test build, /init "should interview you and help setup skills, hooks, etc," which is a concrete shift from static config generation toward guided repo-specific setup new /init test.
The activation path is explicit: launch Claude Code with the environment flag, then run /init inside the target repository. Anthropic's thread details say the command can initialize both fresh and existing repos and set up Claude.md, hooks, and skills, which makes this more of a configuration bootstrapper than a one-off wizard.
Anthropic is framing this as a live test driven by real operator pain points, not a finished launch. The original announcement says the new flow is "based on your feedback," and a follow-up adds that it was informed by "a ton of customer feedback" feedback note customer feedback.
That matters for engineers because /init sits at the boundary between ad hoc prompting and repeatable repo configuration. If the interview flow reliably writes the same Claude.md, hooks, and skills setup Anthropic describes in the thread details, it could reduce the manual glue work teams currently do when standardizing agent behavior across repositories.
The immediate caveat is that skills behavior still looks inconsistent across Claude surfaces. Simon Willison said Claude chat now has an "add to skills" mechanism, but "Claude Code for web does not" as far as he could tell; he surfaced that while experimenting with Starlette 1.0 and Claude skills Starlette skills post Starlette write-up.
A separate practitioner report points to another control problem in the web product: Ethan Mollick asked whether the frontend design skill could be turned off, saying some default Anthropic skills appear to be "always available," which can make it harder to get custom skills selected for overlapping tasks default skill issue. His later follow-up offered no workaround beyond "apparently not" no toggle found. Together, those reports suggest the new /init setup may help teams author skills faster, but not yet guarantee uniform skills behavior across chat, web, and code surfaces.
Vercel Emulate added a programmatic API for creating, resetting, and closing local GitHub, Vercel, and Google emulators inside automated tests. That makes deterministic integration tests easier to wire into CI and agent loops without manual setup.
releaseOpenClaw shipped version 2026.3.22 with ClawHub, OpenShell plus SSH sandboxes, side-question flows, and more search and model options, then followed with a 2026.3.23 patch. Teams get a broader plugin surface, but should patch quickly and review plugin trust boundaries as the ecosystem grows.
releaseCursor shipped Instant Grep, a local regex index built from n-grams, inverted indexes, and Bloom filters that drops large-repo searches from seconds to milliseconds. Faster candidate retrieval shortens the coding-agent loop, especially when ripgrep-style scans become the bottleneck.
breakingChatGPT now saves uploaded and generated files into an account-level Library that can be reused across conversations from the web sidebar or recent-files picker. It removes repetitive re-uploading and makes past PDFs, spreadsheets, and images part of a persistent working context.
breakingEpoch AI says GPT-5.4 Pro elicited a publishable solution to one 2019 conjecture in its FrontierMath Open Problems set, with a formal writeup planned. Treat it as an early milestone worth reproducing, not blanket evidence that frontier models can already automate math research.
I want to make /init more useful- what do you think it should do to help setup Claude Code in a repo?
to give a bit more info: /init helps setup your Claude Code config in a new or existing repo (Claude.MD, hooks, skills, etc) once you start Claude with the flag above, you just need to run /init in the repo of your choice
It's a bit confusing that Claude chat has an "add to skills" mechanism now that makes new skills available when you are using the chat interface but, as far as I can tell, Claude Code for web does not - cc @claudeai did I miss something?
thanks to @IsabellaKHe and @karan_sampath for driving this! it's been informed by a ton of customer feedback that they're very close to
