Scheduling Capability Split Across Claude Code And Cowork (And Its Local Availability Limits)

Issue 56 Edition 2026-02-25 7 min read

General

Sources: 1 • Confidence: High • Updated: 2026-04-12 10:08

Key takeaways

Claude Code does not have a documented mechanism for running tasks on a schedule.
Claude Code supports starting a remote control session on a computer and sending prompts to it from Claude Code web interfaces including web, iOS, and the native desktop app.
A user may see an error that remote control is not enabled for their account even when they are their own administrator, and logging out and back into the Claude Code terminal app can resolve it.
A Claude Code remote control session can encounter a failure mode where every prompt returns an API 500 error.
An author hopes Anthropic is developing a Cowork Cloud product.

Claude Code does not have a documented mechanism for running tasks on a schedule.
Anthropic announced scheduled recurring tasks in Cowork, described as Claude Code's general agent sibling.
Cowork scheduled tasks run only while the computer is awake and the Claude Desktop app is open.
If a Cowork scheduled task run is missed due to sleep or the app being closed, it is skipped and then executed automatically once the computer wakes or the app is opened again.

Claude Code supports starting a remote control session on a computer and sending prompts to it from Claude Code web interfaces including web, iOS, and the native desktop app.
A Claude Code remote control session can be started from the command line using the command "claude remote-control".
After upgrading the Claude iOS app, a remote session can appear in the app as "Remote Control Session (Mac)" under the Code tab.

A user may see an error that remote control is not enabled for their account even when they are their own administrator, and logging out and back into the Claude Code terminal app can resolve it.
Only one Claude Code remote control session can run on a machine at a time.

A Claude Code remote control session can encounter a failure mode where every prompt returns an API 500 error.
If the program on the host machine is restarted during a remote control setup, the existing remote session may return unclear API errors rather than explicitly indicating the session has terminated.

An author hopes Anthropic is developing a Cowork Cloud product.
An author expects current Claude Code remote control issues to be fixed relatively quickly.

Is Claude Code remote control generally available across accounts/plans, and what specific entitlement conditions govern access?
What is the underlying architecture of Claude Code remote control (session routing, authentication model, and how the host machine is addressed), and what failure/timeout semantics are defined?
Does the one-session-per-machine constraint have exceptions (per-user, per-project, per-login) or configurable limits?
Is the "--dangerously-skip-permissions" flag intentionally unsupported in remote control sessions, or is it a bug/oversight, and is there a documented alternative?
How prevalent and persistent are the reported API 500 errors in remote control sessions, and are they correlated with specific actions, clients, or versions?

Scheduling appears fragmented: Cowork adds recurring tasks but depends on a local machine being awake and the desktop app open. This suggests unmet demand for always-on scheduling that could drive interest in a cloud-hosted alternative.
Claude Code remote control enables cross-device prompting into a specific host machine session. This implies a push toward distributed workflows, but current activation friction and one-session limits may constrain multi-user or team adoption.
Operational reliability issues such as entitlement errors requiring re-login and persistent API 500 failures indicate early-stage stability risks. If resolved, usage could expand; if persistent, it may suppress retention and willingness to rely on remote sessions.

Product updates that introduce documented scheduling in Claude Code or add an always-on scheduling mode not dependent on a local awake machine, reducing the current local availability limitation.
Improvements to remote control access and session management such as clearer entitlement gating, fewer re-login fixes, explicit session termination after host restarts, or configurable concurrency beyond one session per machine.
Reports of materially improved remote control reliability, specifically reduced or eliminated API 500 error incidents across clients and versions, with clearer failure and timeout semantics.

Scheduling remains only in Cowork and continues to require a local machine and open desktop app, with no progress toward more reliable always-on execution.
Remote control remains limited to one session per machine with recurring access friction, suggesting the feature is not being scaled for broader team usage.
API 500 failures and unclear session lifecycle behavior persist over time, indicating structural reliability issues that prevent dependable remote control workflows.