Anthropic is transforming how developers handle complex coding tasks with a revamped Projects feature in Claude Code that allows users to run multiple AI agents simultaneously. By introducing a shared memory and centralized coordination, the update enables a team of AI agents to work in parallel on different branches of a repository without losing context. This shifts the AI from a simple chat assistant into a collaborative, multi-threaded development environment.
The upgraded Projects feature operates similarly to other multi-agent frameworks, utilizing a "coordinator" to manage various "threads." Each thread functions as an independent cloud session, tackling specific assignments while sharing a unified library of files, goals, and artifacts. Users can monitor the entire operation through a main project chat or interact with individual threads to fine-tune specific tasks.
Under the hood, each thread is a Claude Code cloud session working on its own branch and copy of the repo. The coordinator keeps work organized, but if any threads work on the same code, the overlap is resolved as a merge conflict just like any other PR. Each thread can further split its delegated work into pieces using subagents, loops, and workflows when needed so large assignments finish faster.
- Anthropic
Availability and Future Local Support
The beta for the redesigned Projects feature is currently live for select Claude Pro and Max subscribers. Anthropic plans to expand access to all Pro, Max, Team, and Enterprise tiers, alongside Cowork and standard Claude chats, in a future rollout.
While threads currently operate exclusively in the cloud, the company confirmed that support for local tools and code execution is arriving "very soon." This upcoming local integration will be critical for developers who need agents to interact directly with their local development environments and proprietary codebases.
The Shift Toward Autonomous Development Teams
Treating AI agents like human developers who generate standard merge conflicts is a massive step toward autonomous software engineering. Instead of relying on a single LLM prompt to generate a monolithic block of code, Claude Code is mimicking a real engineering team's Git workflow. This compartmentalization reduces the risk of AI hallucinating over an entire codebase, as tasks are broken down into subagents and loops with specific boundaries.
If Anthropic successfully implements the promised local tool support, the implications for enterprise development are profound. It could allow these agent swarms to test, debug, and compile code directly on a developer's machine before submitting a pull request. This moves the industry closer to a future where human engineers act primarily as reviewers and architects, while AI agents handle the bulk of the granular coding and conflict resolution.
