Claude Academy

Beyond Local Loops

/loop is great for monitoring during a work session. But it dies when your machine sleeps. /schedule solves this — it creates cloud-based triggers that run independently of your machine.

Think of /schedule as cron jobs powered by Claude Code, running in the cloud.

How /schedule Works

You: /schedule → describe task and cron pattern ↓ Cloud creates a scheduled trigger ↓ At each scheduled time: 1. Cloud spins up a Claude Code session 2. Runs your task against your repository 3. Takes any configured actions (create issues, send notifications) 4. Session ends ↓

Persists indefinitely until you remove it

The key difference from /loop: your machine doesn't need to be on. The schedule runs in the cloud.

Creating a Schedule

/schedule "Every day at 9 AM UTC, check for dependency vulnerabilities: 1. Run pnpm audit 2. If critical or high vulnerabilities found, create a GitHub issue with the details

3. If no vulnerabilities, log success"

Claude creates the schedule:

Schedule Created: dependency-audit Cron: 0 9 * Next run: 2026-04-05 09:00 UTC

Task: Check dependency vulnerabilities

Cron Syntax Reference

┌───── minute (0-59)
 │ ┌───── hour (0-23)
 │ │ ┌───── day of month (1-31)
 │ │ │ ┌───── month (1-12)
 │ │ │ │ ┌───── day of week (0-7, 0 and 7 = Sunday)
 │ │ │ │ │
     *

Common patterns:

| Cron | Meaning |

|---|---|

| 0 9 * | Every day at 9:00 AM |

| 0 9 1 | Every Monday at 9:00 AM |

| 0 9 1-5 | Weekdays at 9:00 AM |

| 0 /4 | Every 4 hours |

| 0 0 1 | First of every month at midnight |

| 0 9,17 1-5 | 9 AM and 5 PM on weekdays |

Practical Scheduled Tasks

Daily Dependency Audit

/schedule "Daily at 9 AM UTC (0 9 *): Check for dependency vulnerabilities with pnpm audit. If critical vulnerabilities found: - Create a GitHub issue titled 'Critical Dependency Vulnerability — [date]' - Include the affected packages and recommended fixes - Label it 'security' and 'priority-high'

If no vulnerabilities: log 'All clear' and take no action."

Weekly Security Scan

/schedule "Every Monday at 8 AM UTC (0 8 1): Run a comprehensive security scan on the codebase: 1. Check all files in src/ for common vulnerabilities 2. Look for hardcoded secrets, SQL injection, XSS patterns 3. Check for new dependencies added since last week 4. Generate a security report as a GitHub issue

5. Tag it 'weekly-security-review'"

Daily Test Suite Monitor

/schedule "Every day at 6 AM UTC (0 6 *): 1. Pull the latest main branch 2. Run pnpm test 3. If any tests fail: - Create a GitHub issue with failing test details - Label it 'broken-tests' and 'priority-high'

4. If all pass, do nothing."

Weekly Report Generation

/schedule "Every Friday at 5 PM UTC (0 17 5): Generate a weekly project report: 1. Count commits this week (git log --since='1 week ago') 2. List PRs merged 3. List issues closed 4. Check test coverage

5. Create a summary and post it as a GitHub discussion"

Monthly Dependency Update

/schedule "First Monday of each month at 10 AM UTC (0 10 1-7 * 1): 1. Check for outdated dependencies (pnpm outdated) 2. For patch updates: create a PR that updates them 3. For minor updates: create an issue listing them 4. For major updates: create an issue with migration notes

5. Run tests after any updates"

Managing Schedules

List All Schedules

# In session /schedule list # From terminal

claude schedule list

Output:

Active Schedules Name │ Cron │ Next Run │ Last Result ────────────────────┼───────────────┼────────────────────┼───────────── dependency-audit │ 0 9 * │ 2026-04-05 09:00 │ ✓ No issues security-scan │ 0 8 1 │ 2026-04-07 08:00 │ ✓ 0 findings test-monitor │ 0 6 * │ 2026-04-05 06:00 │ ✗ 2 failures

weekly-report │ 0 17 5 │ 2026-04-11 17:00 │ ✓ Posted

View Schedule Details

/schedule show dependency-audit

Update a Schedule

/schedule update dependency-audit

"Change the cron to run twice daily: 9 AM and 5 PM (0 9,17 *)"

Delete a Schedule

/schedule remove dependency-audit

Run Immediately

/schedule run dependency-audit

Triggers an immediate execution without waiting for the next scheduled time.

/loop vs. /schedule

| Feature | /loop | /schedule |

|---|---|---|

| Runs on | Your machine (local process) | Cloud infrastructure |

| Survives reboot | No | Yes |

| Survives sleep | No | Yes |

| Interval | Minutes (5m, 30m, 2h) | Cron schedule |

| Best for | Active monitoring during work | Recurring automation |

| Cost | Your session tokens | Cloud execution tokens |

| Interactive | Yes (you can steer) | No (runs autonomously) |

Use /loop when you're actively working and want to monitor something. Use /schedule when you want a task to run reliably regardless of whether you're at your computer.

Combining Schedules with Other Features

Schedule + MCP

/schedule "Daily at 9 AM: Check GitHub for PRs older than 48 hours without review.

For each stale PR, post a comment reminding the author and tag the team lead."

Schedule + Commands

/schedule "Weekly on Monday: run /security-review on all changes since last Monday.

Post the results as a GitHub discussion."

Schedule + Notifications

/schedule "If any scheduled task fails, send a Slack notification to #engineering-alerts

with the task name and error details."

Key Takeaway

/schedule creates cloud-based cron jobs powered by Claude Code. Unlike /loop (which runs locally and dies when your machine sleeps), /schedule persists independently. Use it for daily dependency audits, weekly security scans, test suite monitoring, and automated report generation. Manage schedules with /schedule list, show, update, remove, and run. Choose /loop for active monitoring during work sessions and /schedule for reliable recurring automation.

Cloud Automation with /schedule

Learning Objectives