Skip to main content
Prevent tasks from running indefinitely by setting a per-project timeout. When a task exceeds the limit, it’s automatically terminated and follows the retry policy.

​
Configuration

In Project Settings > Automation, set Task Timeout in minutes:
ValueBehavior
0 (default)No timeout — tasks run until completion
1–1440Auto-kill after this many minutes (max 24 hours)
Start with a generous timeout (30–60 minutes) and adjust down based on your typical task duration. Check the Analytics page to see average task times.

​
How It Works

Task starts

A timer begins when the agent process spawns.

Monitoring

The queue poll thread checks active processes every 15 seconds against the timeout.

Timeout triggered

When a task exceeds the limit:
  1. Process is terminated (SIGTERM on macOS/Linux, taskkill on Windows)
  2. Task follows the retry policy — retry with backoff, or move to Failed
  3. Webhook notification fires (task_timeout event)
  4. Timeout logged in task history

​
Use Cases

Rate Limit Loops

Kill tasks stuck in rate limit wait cycles that would otherwise run forever.

Infinite Loops

Terminate tasks where Claude enters a reasoning or editing loop.

Resource Management

Ensure no single task monopolizes an agent slot, keeping the queue healthy.
A timed-out task counts as a failure and consumes a retry attempt. If all retries are exhausted, the task moves to Failed status.