[FEATURE] Native automations with scheduled tasks and review queue

1. TITLE

Native automations with scheduled tasks and review queue

2. PREFLIGHT CHECKLIST

• [ ] I have searched existing requests and this feature hasn't been requested yet
• [ ] This is a single feature request (not multiple features)

3. PROBLEM STATEMENT

I want to run recurring Claude Code tasks automatically (daily code reviews, issue triage, CI failure summaries, dependency updates) without manual intervention. Currently, this requires:

1. Setting up external schedulers (cron, GitHub Actions)
2. Writing wrapper scripts to invoke Claude CLI
3. Parsing output and routing results manually
4. No unified interface to review what automations produced

The friction is high enough that I don't automate tasks I should. When I do automate, results are scattered across logs, emails, and CI artifacts with no coherent review workflow.

4. PROPOSED SOLUTION

Add a native Automations system to Claude Code:

Creating Automations:

/automation create "Daily issue triage"
  --schedule "0 9 * * *"  # 9am daily
  --skill issue-triage
  --prompt "Review new GitHub issues, categorize by severity, suggest assignments"

Or via Desktop UI: Automations panel → "+ New Automation" with form fields.

Automation Components:

• Instructions: What Claude should do
• Skills (optional): Pre-attach skills like github, linear, slack
• Schedule: Cron expression or presets (daily, weekly, hourly)
• Output destination: Review queue, file, webhook, or notification

Review Queue:

• Dedicated tab/panel in Desktop app showing automation results
• Each result is resumable - click to continue conversation
• Mark as reviewed, snooze, or re-run
• Filter by automation name, date, status

Management:

• /automation list - Show all automations with last run status
• /automation pause <name> - Temporarily disable
• /automation logs <name> - View execution history

5. ALTERNATIVE SOLUTIONS

Current workarounds:

1. GitHub Actions + Claude CLI: Works but no review queue, results scattered in workflow logs
2. Cron + shell scripts: No Desktop integration, manual result parsing
3. Manual daily routine: Time-consuming, easy to forget

What Codex does:
"With the Codex app, you can also set up Automations that let Codex work in the background on an automatic schedule. Automations combine instructions with optional skills, running on a schedule you define. When an Automation finishes, the results land in a review queue so you can jump back in and continue working if needed."

This is exactly what I need - scheduled execution with a unified review interface.

6. PRIORITY

• [ ] Critical - Blocking my work
• [x] High - Significant impact on productivity
• [ ] Medium - Would be very helpful
• [ ] Low - Nice to have

7. FEATURE CATEGORY

• [ ] CLI commands and flags
• [ ] Interactive mode (TUI)
• [ ] File operations
• [ ] API and model interactions
• [ ] MCP server integration
• [ ] Performance and speed
• [ ] Configuration and settings
• [x] Developer tools/SDK
• [ ] Documentation
• [ ] Other

8. USE CASE EXAMPLE

Scenario: Automated daily development workflow

1. I create three automations:

Morning Issue Triage (9am daily):

• Review new GitHub issues from last 24 hours
• Categorize: bug, feature, question, invalid
• Suggest priority and assignee
• Draft response templates for common questions

CI Failure Summary (on failure):

• Parse failed CI logs
• Identify root cause
• Suggest fix or link to similar past failures
• Optionally create draft PR with fix

Weekly Dependency Audit (Monday 10am):

• Check for outdated dependencies
• Review changelogs for breaking changes
• Create PR with safe updates
• Flag risky updates for manual review

1. Each morning, I open Claude Code Desktop and check the Review Queue:

• "Issue Triage - 5 new issues categorized" → Review assignments, approve
• "CI Failure - test_auth.py line 42 TypeError" → Click to see suggested fix, apply it
• "Dependency Audit - 3 updates ready" → Review PR, merge safe ones

1. Click any result to continue the conversation with full context preserved.

Time saved: ~1 hour daily of manual triage and monitoring.

9. ADDITIONAL CONTEXT

Technical considerations:

• Could run locally (when Desktop is open) or in cloud (always-on)
• Claude Code hooks already support event-driven automation; this extends to time-driven
• Session resume capability means automation results can become full conversations
• Rate limiting: respect API quotas, queue automations if needed

Related features:

• Webhook triggers (on PR, on issue, on Slack message)
• Automation templates for common workflows
• Team-shared automations via .claude/automations/

Prior art:

• Codex Automations
• GitHub Actions scheduled workflows
• Zapier/IFTTT for non-code automation