claude-auto-resume

https://github.com/terryso/claude-auto-resume

[{"step":"Identify the paused task","action":"Run `claude --list-paused` to find the task ID and command. Note the exact command used to initiate the task (e.g., `claude --analyze dataset.csv --model haiku-3.5`).","tip":"Use `claude --verbose` for detailed logs to confirm the task was paused due to usage limits."},{"step":"Set up automation","action":"Create a shell script (e.g., `resume_claude_task.sh`) that includes the paused task's command and resume logic. Example: `#!/bin/bash\nTASK_ID=\"cli_7f8a2b1c\"\nLOG_FILE=\"/var/log/claude/resume_${TASK_ID}_$(date +%Y%m%d).log\"\nclaude --resume $TASK_ID --continue >> $LOG_FILE 2>&1\n# Add notification logic (e.g., curl for Slack webhook)","tip":"Test the script manually first to ensure the resume command works as expected."},{"step":"Schedule periodic checks","action":"Use `cron` to run the script every 5-15 minutes during business hours. Example: `*/10 * * * * /path/to/resume_claude_task.sh`","tip":"Adjust the frequency based on your usage limits reset time (e.g., every 5 minutes if limits reset hourly)."},{"step":"Monitor and log results","action":"Check the log file for errors or completion status. Use `claude --status [TASK_ID]` to verify progress. Set up alerts for failures (e.g., `if ! grep -q \"completed\" $LOG_FILE; then send_alert; fi`).","tip":"Use tools like `logrotate` to manage log file sizes if running long-term."},{"step":"Handle edge cases","action":"Add fallback logic for tasks that fail to resume automatically. Example: `if ! claude --status $TASK_ID | grep -q \"running\"; then claude --force-resume $TASK_ID; fi`","tip":"Test force-resume scenarios in a non-production environment first."}]