Anthropic Launches Claude Code Projects for Developers
Anthropic has redesigned Claude Code Projects as a persistent workspace for long-running software development.
The new experience uses a central Claude conversation to coordinate multiple coding sessions. As a result, developers can describe a broader goal while Claude divides the work into parallel threads.
Claude Coordinates Parallel Development Threads
Claude Code Projects introduces a coordinator above individual Claude Code sessions. Developers can continue briefing this coordinator as project requirements change. Meanwhile, Claude can assign new requests to existing threads or create additional ones.
Each worker thread runs as a Claude Code cloud session. It also works from its own branch and repository copy. As a result, several development tasks can progress at the same time.
For example, a developer could ask Claude to reduce an application’s checkout latency. Claude could then create separate threads for profiling, testing optimisations, and preparing pull requests.
Similarly, teams could connect API, web, and mobile repositories. Claude could then divide a deprecated API migration across those repositories while tracking dependencies between the changes.
However, parallel work does not remove normal software development issues. If multiple threads change overlapping code, developers can still encounter conventional merge conflicts.
Shared Memory Keeps Projects Moving
A central feature is shared Project memory. Every worker thread can contribute to and use that memory as development continues. As a result, Claude can retain decisions, requirements, and other project information across sessions.
The Project also includes a library for user-provided files and artefacts created during development. Users can also configure project environments, connectors, plugins, instructions, and model settings.
This approach differs from Anthropic’s existing Claude Projects feature. The earlier product organises chats, files, instructions, and project knowledge. By contrast, Claude Code Projects focuses on coordinating ongoing development work across multiple coding sessions.
This distinction matters for teams running development over weeks or months. Instead of reconstructing context at the start of every session, developers can keep one continuing project conversation.
Furthermore, Anthropic says developers can adjust how frequently Claude checks in, how often it creates threads, and how detailed its updates should be. Users can also inspect individual threads when they need to review or redirect work.
Beta Begins With Select Pro and Max Users
The Claude Code Projects beta initially targets select Claude Pro and Max subscribers using cloud sessions in Claude Code. Users who already have Projects on the web or desktop are excluded from this initial group.
Anthropic plans to expand access to additional Claude Code users on those plans. Eventually, the redesigned experience is expected to reach more Claude users, including Team and Enterprise customers.
For now, worker threads run in the cloud. That lets them keep working after developers close their laptops. However, they cannot yet access files or tools on a developer’s local computer or internal network. Anthropic says local support is coming soon.
Meanwhile, Claude Code Projects uses existing subscriber usage rather than a separate pricing model. Because every worker thread counts as a full Claude Code session, running several threads can use up usage limits faster.
The launch reflects a broader shift toward persistent AI coding workflows. Anthropic has already added subagents, hooks, background tasks, checkpoints, and cloud-based development capabilities to Claude Code.
Now, Claude Code Projects adds a coordination layer on top of those capabilities. The result is a development workflow in which one ongoing conversation tracks the broader objective while multiple AI sessions handle individual tasks.
