Anthropic rebuilds Claude Code Projects around parallel agent threads

single source· 1 articles · confidence: low · first seen 2026-09-17 18:58 UTC

What this means for you

Nothing to act on yet: no availability date, no thread limits, no pricing for the shared-memory layer. If you hand-roll multi-agent pipelines in Claude Code — separate sessions passing notes to each other — this is the thing to watch, because shared memory inside one project is what would change token cost and duplication. Wait for the details.

Anthropic has rebuilt the Projects feature in Claude Code so that a single project can hold several agents at once, The Verge reports. Each project keeps a shared memory, shared goals and a common library of files and artefacts; inside it, "threads" run separate tasks in parallel while a coordinator directs the work. The design resembles a small group of tools that manage agents in groups rather than one at a time, which The Verge illustrates with Grok Bot. No pricing, availability date or evaluation figures are given, and this account rests on that single short report.

Key facts

  • ·Anthropic's revamped Claude Code Projects lets multiple agents run within one project with shared memory, goals and a library of files and artefacts. source
  • ·Each project contains "threads" that run different tasks in parallel. source
  • ·A coordinator directs the threads within a project. source
  • ·The Verge compares the design to Grok Bot and other tools that manage groups of agents. source
  • ·No price, availability date or performance figures for the relaunched Projects feature are reported. source

What the sources say

  • The Verge AI — Short report on the feature's structure: shared memory, parallel threads and a directing coordinator.

Sources

The original reporting. Follow these — they did the work.

Related stories

← the wire