agent-limit-retry

remote_copper_carolynn

Keep Claude Code working through usage limits: exact reset times, handoff, auto-resume.

1likes0comments

Visit project 1

Project details

About this project

On a Claude Pro/Max plan the 5-hour and weekly windows close mid-task. agent-limit-retry reads what Claude Code already writes locally (status-line rate_limits, exact reset times in transcripts), asks the agent to write a handoff file at 90%, records the exact reset time on a hit, and lets headless runs sleep through the reset and resume the same session. MIT, no proxy, no telemetry.

What it does

Claude Code's own auto-continue only covers the main conversation, in an interactive session, when the reset is under 24 hours away. It does not cover headless claude -p runs, weekly limits, or the work that was in flight when the limit hit. agent-limit-retry fills those gaps by reading files Claude Code already writes and telling the agent what to do at the right moment.

Key features

Status line tap: saves your 5-hour / 7-day percentages and reset times, keeps your own status line
Handoff before the limit: at 90% of a window the agent finishes its step and writes HANDOFF.md
Exact reset time on a limit hit, plus a desktop notification
alr run "<task>": headless runs that sleep until the reset and --resume the same session (weekly limits too)
alr selftest: simulated limit hit + resume against a local mock API, no quota spent

Install

npx agent-limit-retry
alr selftest

Node.js 18+, Claude Code 2.1.234+. Everything runs locally. Source and docs: https://github.com/error0702/agent-limit-retry

A paid Pro build adds resuming the subagents a limit killed (by id, after the reset, for Claude Code / Codex / ZCode): https://retry.autorun.fun