Dashboard-Driven Development

This scenario shows how a product trio (PM, designer, tech lead) drives delivery by treating the Spec Kitty kanban dashboard as the single source of truth.

Setup

Start the dashboard in your project:

cd my-project
spec-kitty dashboard
# Dashboard opens at http://localhost:3000 (or next available port)
# Runs in background, auto-starts on system reboot

Team Roles

  • PM watches the dashboard overview tab in browser, resolving blockers and rebalancing priorities
  • Designer reviews artifacts linked in the feature cards (spec.md, plan.md, quickstart.md) to ensure UX intent stays intact
  • Tech Lead keeps an eye on lane distribution and redirects agents if review queues build up

Daily Loop

1. Morning Alignment (9:00 AM)

  • All: Open dashboard at http://localhost:3000
  • PM: Reviews lane counts per feature, flags work packages stuck in "doing" >24 hours
  • Tech Lead: Notes which agents are assigned to which tasks (shown in dashboard)

2. Assign Work Packages (9:30 AM)

  • Tech Lead: Runs .kittify/scripts/bash/tasks-list-lanes.sh FEATURE-SLUG to spot idle prompts
  • PM: Assigns owners via Slack/email based on dashboard view
  • Agents: Move prompts to "doing": 
    spec-kitty agent action implement WP01

3. Midday Review (1:00 PM)

  • Designer: Checks prompts in for_review/ lane (visible in dashboard)
  • Designer: Opens work package files, adds feedback as markdown comments
  • Designer: Moves back to planned or forward to done: 
    # If needs rework:
spec-kitty agent action review WP01

# If approved:
spec-kitty agent action implement WP01

4. Evening Recap (5:00 PM)

  • PM: Takes dashboard screenshot showing lane distribution
  • PM: Exports tasks.md for async updates to stakeholders
  • Tech Lead: Uses dashboard API for metrics: 
    curl http://localhost:3000/api/features | jq

Dashboard Features Used

  • Real-time updates - Browser auto-refreshes as agents move tasks
  • Lane visualization - See work distribution: planned (blue), doing (yellow), review (orange), done (green)
  • Agent assignments - Each work package shows which agent is handling it
  • Completion metrics - Progress bars show % complete per feature
  • Artifact links - Click through to spec.md, plan.md from dashboard

Tips

  • Use dashboard search/filter to focus on a single feature when managing multiple
  • Dashboard runs in background - access from any browser on the network
  • /spec-kitty.dashboard command from any agent reopens the dashboard URL
  • Refresh is automatic via polling - no manual page reload needed
  • Pair dashboard with docs/kanban-dashboard-guide.md for advanced features
Edit this page