BotBeat
...
← Back

> ▌

AnthropicAnthropic
PRODUCT LAUNCHAnthropic2026-03-07

Claude Code Introduces Scheduled Tasks for Automated Prompt Execution

Key Takeaways

  • ▸Claude Code now supports automated prompt execution through scheduled tasks using the /loop command and cron-style scheduling
  • ▸Tasks can be recurring (with flexible interval syntax) or one-time reminders, parsed from natural language input
  • ▸Scheduling is session-scoped only—tasks don't survive process restarts, though durable alternatives exist for persistent automation
Source:
Hacker Newshttps://code.claude.com/docs/en/scheduled-tasks↗

Summary

Anthropic has expanded Claude Code's capabilities with a new scheduled tasks feature that allows users to run prompts automatically on recurring intervals or as one-time reminders. The feature, accessible through the /loop command and cron scheduling tools, enables developers to automate routine checks such as monitoring deployments, tracking pull requests, or following up on long-running builds without manual intervention.

The scheduling system supports flexible interval syntax, accepting time units in seconds, minutes, hours, or days, with natural language parsing that converts user input into cron expressions. Users can schedule recurring tasks with commands like '/loop 5m check if the deployment finished' or set one-time reminders with natural language requests such as 'remind me at 3pm to push the release branch.' Tasks can also wrap other commands, allowing developers to repeatedly execute pre-packaged workflows.

Scheduled tasks are session-scoped, meaning they persist only within the current Claude Code process and terminate when the session ends. The system includes built-in jitter to prevent API overload, adding small deterministic offsets to fire times—up to 10% of the period for recurring tasks (capped at 15 minutes) and up to 90 seconds for one-time tasks scheduled at common intervals. Sessions can maintain up to 50 scheduled tasks simultaneously, with tasks automatically expiring after three days to prevent resource accumulation.

  • Built-in jitter prevents API congestion by adding small, deterministic delays to scheduled execution times
  • Sessions support up to 50 concurrent scheduled tasks, each with a three-day automatic expiry

Editorial Opinion

This feature represents a thoughtful expansion of Claude Code's developer tooling, addressing a common workflow pain point—the need to manually check on long-running processes. The session-scoped limitation is both a strength and weakness: it prevents runaway resource consumption and keeps the feature simple, but also means developers can't rely on it for truly persistent automation. The inclusion of jitter to prevent thundering herd problems shows production-grade engineering thinking, though the three-day expiry feels somewhat arbitrary and could frustrate users running extended development sessions.

Large Language Models (LLMs)AI AgentsMLOps & InfrastructureProduct Launch

More from Anthropic

AnthropicAnthropic
RESEARCH

Inside Claude Code's Dynamic System Prompt Architecture: Anthropic's Complex Context Engineering Revealed

2026-04-05
AnthropicAnthropic
POLICY & REGULATION

Anthropic Explores AI's Role in Autonomous Weapons Policy with Pentagon Discussion

2026-04-05
AnthropicAnthropic
POLICY & REGULATION

Security Researcher Exposes Critical Infrastructure After Following Claude's Configuration Advice Without Authentication

2026-04-05

Comments

Suggested

AnthropicAnthropic
RESEARCH

Inside Claude Code's Dynamic System Prompt Architecture: Anthropic's Complex Context Engineering Revealed

2026-04-05
OracleOracle
POLICY & REGULATION

AI Agents Promise to 'Run the Business'—But Who's Liable When Things Go Wrong?

2026-04-05
Google / AlphabetGoogle / Alphabet
RESEARCH

Deep Dive: Optimizing Sharded Matrix Multiplication on TPU with Pallas

2026-04-05
← Back to news
© 2026 BotBeat
AboutPrivacy PolicyTerms of ServiceContact Us