Breaking

Claude Code Adds Graceful Wrap-Up for Tasks That Hit the 5-Hour Limit

September 25, 2026 at 15:06 EDT

Software Dev & Coding
AI Agents

Claude Code will now try to end a task at a graceful stopping point when a user hits the 5-hour limit mid-run, using a small fixed allowance deducted from the weekly limit to wrap up what it can. The change is a reliability update for long agentic coding sessions, not a quota increase.

Sep. 25, 2026 · Claude Code

A softer stop for long coding runs

Claude Code now tries to find a clean stopping point when a running task reaches the 5-hour usage window, using a small fixed allowance from the weekly limit to wrap up what it can.

What changes at the limit

5 hours

short-window trigger

fixed

wrap-up allowance

weekly

where the allowance is deducted

Usage scale still depends on plan

Max tiers remain larger than Pro; the wrap-up behavior does not expand the plan ceiling.

New end-of-window flow

Limit arrives

A running task reaches the session window.

→

Find a stop

Claude looks for a graceful stopping point.

→

Use allowance

A small fixed amount is pulled from weekly usage.

→

Wrap what it can

The task gets a cleaner handoff before stopping.

Why developers wanted it

Long agentic runs can otherwise stop mid-edit, mid-subagent, or before a validation pass. A cleaner stop should reduce recovery work.

What it does not change

The allowance size is undisclosed, and the weekly budget still pays for the wrap-up. This is reliability, not extra quota.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.