Anthropic Launches Claude Code Projects for Persistent Dev Orchestration

Anthropic has launched Claude Code Projects, an always-on AI coordination agent designed to remember and delegate persistent development work. [3] The tool allows users to describe their work in one conversation, which the system then subdivides and manages across parallel threads. [2][3] As the work progresses, Claude Code Projects automatically coordinates these threads and maintains a shared memory of the project's evolution. [3] Previously, developers continuing work across multiple sessions had to manually maintain an external record of their status and feed it back into the harness. [3] The update is specifically targeted at enterprise applications, which require continual patching and adaptation over years. [3]

Anthropic’s update shifts Claude from a stateless coding assistant to a persistent orchestrator of ongoing development cycles. By maintaining project context and delegating tasks across parallel threads, Claude Code Projects directly addresses the friction of multi-session workflow management for enterprise developers. This evolution positions Claude more aggressively against other AI coding agents by emphasizing long-term orchestration over single-shot code generation. What the sources don't address: How the shared memory limits scale over time or what the pricing structure is for these always-on coordination agents.