· Updated
Monitor tool for Claude Code, its terminal-based agentic coding tool. The feature allows the agent to execute background scripts that wake the model only when specific conditions are met. This replaces polling—repeatedly checking a status at set intervals—which often wastes tokens in long-running agent loops.This update addresses the high cost and context waste of autonomous agents. By offloading observation to lightweight scripts, the agent remains dormant until an event requires attention. It transforms the agent into a persistent observer that only consumes tokens when actively processing a trigger rather than during idle waiting periods.
You can now prompt Claude Code to monitor external processes or logs in real-time. Once a background script detects a trigger, it streams the output back into the conversation for immediate action. This is available now for developers using the Claude Code preview to manage asynchronous DevOps tasks or long-running test suites.
Thrilled to announce the Monitor tool which lets Claude create background scripts that wake the agent up when needed. Big token saver and great way to move away from polling in the agent loop Claude can now: * Follow logs for errors * Poll PRs via script * and more! https://t.co/eflixzi0xk
