Anthropic upgrades Claude Code into a full‑workstation desktop app for autonomous

Anthropic’s overhaul of Claude Code arrives just as developers are demanding more “always‑on” assistants that can juggle dozens of tasks without a human constantly shepherding each one. The new desktop client, announced on April 15, replaces the single‑session interface with a full‑workstation layout that lets users spin up multiple autonomous coding agents in parallel, according to the Abz Global report [Abz]. A sidebar now lists every active and recent run, filterable by status, project, or environment, and automatically archives sessions when a pull request merges or closes. This visual overhaul is more than cosmetic; it reflects Anthropic’s belief that serious Claude Code users already operate several agents at once and need a single pane of glass to keep them in sync.

Beyond the UI, the app embeds core development tools that were previously external. An integrated terminal and in‑app file editor let developers run tests, make spot edits, and ship code without hopping between VS Code, iTerm or other shells, the same Abz article notes. The diff viewer has been rebuilt to handle large changesets, while the preview pane now renders HTML, PDFs, and even local app servers, turning the desktop client into a self‑contained development environment. Shortcut‑driven side chats (⌘ + ; or Ctrl + ;) spawn auxiliary conversations that branch off a main session, preserving the primary agent’s context while still allowing ad‑hoc queries—a feature that mirrors the “side chats” described in Damien Gallagher’s BuildRLab piece [Gallagher].

The functional upgrades are tightly coupled with Anthropic’s broader push toward autonomous developer workflows. Gallagher reports that the update introduces repeatable routines, checkpoints, background tasks, hooks, and sub‑agents, all powered by the latest Sonnet 4.5 model. These capabilities let Claude Code execute multi‑step operations—such as running a test suite, committing a change, and opening a pull request—without a human typing each command. The addition of a native VS Code extension further blurs the line between the AI assistant and the developer’s primary IDE, suggesting Anthropic is positioning Claude Code as a “real workhorse” rather than a mere chatbot.

The timing of the workstation launch coincides with Anthropic’s rollout of identity verification for high‑risk Claude API usage, a safety measure outlined in a separate AI‑Tools‑Hub‑Daily post [Verification]. The verification process, which requires multi‑factor authentication and document checks for organizational users, targets enterprise scenarios handling sensitive data—financial services, healthcare, and the like. By tightening access while simultaneously expanding Claude Code’s autonomy, Anthropic appears to be hedging against misuse even as it hands developers more power. The dual strategy underscores a growing industry consensus that AI productivity tools must be both capable and accountable.

Early adopters are already testing the limits of the new workstation. Developers report that the drag‑and‑drop workspace lets them arrange the terminal, diff viewer, editor, and preview pane in any grid that matches their workflow, a flexibility highlighted by both Abz and Gallagher. The ability to run “dozens of parallel coding agents without losing track,” as the original lede promised, is now tangible: a single Claude Code instance can spin up multiple sub‑agents to refactor code, generate documentation, and even spin up local servers, all while the main sidebar keeps a tidy ledger of each task’s status. For teams that need rapid iteration across multiple micro‑services, this could shave hours off the development cycle.

Anthropic’s move signals a shift in the developer‑tool market from “assistant‑on‑demand” to “assistant‑as‑a‑workspace.” By embedding terminal access, file editing, diff visualization, and preview capabilities directly into Claude Code, the company is challenging the dominance of traditional IDE extensions that merely surface AI suggestions. If the workstation lives up to its promise, it could become the de‑facto hub for autonomous coding—especially for enterprises that have already cleared the new identity‑verification hurdle. The real test will be whether developers embrace a single, AI‑driven environment over their entrenched toolchains, but for now, Anthropic has certainly raised the bar for what a coding assistant can look like.