Anthropic has redesigned Claude Code Projects around one coordinating conversation and multiple cloud work threads. The beta can keep related engineering work moving in parallel, but each thread is still a full session with its own usage, branch and review burden.
RELATED BUYER PROFILEReview Claude pricing, limits and independent-test-pending evidencePricing · access · strengths · limitations →RELATED DECISION GUIDEOpen the developer profession decision deskWorkflow · evidence · risk · governance →CONTINUE THE DECISIONRun a same-repository coding pilotEvidence · workflow · next action →CONTINUE THE DECISIONSet repository, secret and approval boundariesEvidence · workflow · next action →CONTINUE THE DECISIONMeasure complete cost per accepted changeEvidence · workflow · next action →What you need to know
- The redesigned Projects experience is initially a beta for selected Claude Pro and Max users who use Claude Code cloud sessions
- Each thread is a separate cloud session with its own copy and branch; overlapping edits can still create merge conflicts
- Several full sessions can consume plan usage faster, so buyers should measure accepted changes and review effort—not the number of threads opened
The verified beta boundary
Short answer: orchestration is the new feature
Anthropic announced the redesigned Projects experience on 17 September 2026. Instead of asking a user to manage a folder of chats, one coordinating conversation can scope a request, delegate work to parallel Claude Code threads, review their outputs and assemble a result. The feature is useful when work can be separated cleanly. It does not remove the need for repository permissions, testing, code review or an accountable merge decision.
Access starts with selected Pro and Max users
The initial beta is for selected Claude Pro and Max users who use Claude Code cloud sessions and do not already have projects on the web or desktop. Anthropic says it plans to expand access to more Claude Code users on those plans over the following week, with Team and Enterprise support later. Existing projects stay in the current experience until upgraded. This is a staged rollout, not universal UK availability or a separate free trial.
There is no separate Projects price
Anthropic's current pricing page lists Pro at $20 monthly or $17 per month when billed annually, and Max from $100 monthly. Claude Code is included, but Projects does not have a separate published price. Several simultaneous threads are still several full Claude Code sessions, and usage across Claude surfaces shares the plan allowance. Buyers should confirm signed-in UK billing, limits and any extra-usage terms before estimating cost.
Each thread has a branch—and overlap becomes merge work
A project can define a cloud environment, connectors, plugins, instructions and model. Every delegated thread runs as its own Claude Code cloud session with its own copy of the repository and its own branch. That isolates work, but Anthropic explicitly notes that changes to the same code can create merge conflicts. Split work by clear ownership boundaries and keep a human responsible for tests, conflict resolution and the final merge.
Shared memory helps context, not correctness
Threads can share project-level memory and a library of project context, while each thread can use subagents, loops and other Claude Code workflows. Shared context may reduce repeated explanation, but it can also spread an incorrect assumption across several workstreams. Store only reviewed project instructions, make acceptance tests explicit and require every thread to report changed files, commands run, tests passed and unresolved risks.
Cloud execution is the current security boundary
The beta currently runs threads in Claude Code cloud sessions. Anthropic says work on local machines and behind a company's network is coming soon, so buyers should not imply those routes already exist. Before connecting a repository, verify organisation approval, identity and repository scope, secrets handling, connector permissions, retention, activity logs, branch protection and the procedure for revoking access.
A useful pilot has one goal and separable workstreams
Choose a bounded task such as reducing one measured performance bottleneck or migrating a defined package. Freeze the repository snapshot, tests and acceptance criteria. Compare the redesigned Project with the current single-session workflow. Record accepted pull requests, human interventions, failed tests, merge conflicts, review minutes, elapsed time and total usage. Parallel execution is valuable only if it reduces complete cost per accepted change.
HubAI buyer verdict
Claude Code Projects is a credible beta for experienced engineering teams that already use Claude Code and can divide work without creating hidden coupling. It is not yet a reason to change an enterprise coding stack: Team and Enterprise access is still later, current execution is cloud-based and independent same-repository evidence is missing. The existing Claude profile remains Test pending and receives no HubAI Score.
Evidence limits and editorial disclosure
Independent editorial coverage; not sponsored. Access, orchestration, pricing and usage statements were checked against Anthropic's official Projects announcement and Claude pricing page on 21 September 2026. HubAI has not independently benchmarked completion, speed, merge quality or cost. The visible UK Google Trends first 25 included only a generic AI cluster during this review, not a Claude Code Projects query, so no story-specific search volume is claimed. The cover is an original conceptual illustration, not a product interface or Anthropic logo.
HUBAI VIEWUse the beta for one bounded project with separable workstreams. Measure accepted changes, merge conflicts, review time and complete usage before treating parallelism as productivity.
Buyer decision signal: Projects beta · parallel cloud threads
What to verify next
1Confirm the account is actually included in the staged Pro or Max beta
2Choose one bounded repository task with written acceptance tests
3Split threads by clear file, package or responsibility boundaries
4Limit repository, connector, secret and cloud-environment permissions
5Require every thread to report changes, commands, tests and unresolved risk
6Measure merge conflicts, review minutes and total usage per accepted change
7Keep branch protection, human approval and a rollback route in place
Read the evidence
Capabilities, availability and prices can change. HubAI keeps analysis separate from the underlying official material.
01Anthropic: Projects redesigned — 17 September 2026Open source ↗02Claude plans and pricingOpen source ↗03Claude Code product overviewOpen source ↗
Code AI
Code AI
Code AI
Code AI