π Quick Start: Continuous Agents
5-Minute Setup
1. Start the System
cd /Users/kbselander/Developer/Notebook/mcp-servers/servers/mcp-server-altsportsleagues.ai/2.1-cloud-run-docker-mcp
# Start your frontend first
cd clients/frontend
npm run dev &
# Go back to scripts
cd ../../scripts/deploy
# Start all agents
./start-continuous-system.sh start2. Watch It Work
# View status
./start-continuous-system.sh status
# Watch logs (Ctrl+C to exit)
./start-continuous-system.sh logs
# View kb.json
./start-continuous-system.sh view3. Make a Change
# Edit any file in clients/frontend
# Save it
# Watch agents detect and fix issues automatically
# Check kb.json to see updates4. Check Deployability
# Run once to check current state
./continuous-frontend-nextjs16-qa-compliance-eval-system.sh once
# If output shows:
# β
Deployable: YES
# Then you're safe to deploy!5. Stop When Done
./start-continuous-system.sh stopWhat's Happening Behind the Scenes
Every 5 minutes: QA Agent tests all pages with 3 ghost clickers
Every 10 minutes: Style Agent checks design system compliance
Every 3 minutes: Improvement Agent reads kb.json and applies fixes
Continuous: All agents learn from fix success/failure rates
On Threshold: System escalates to human if stuckKey Files
- kb.json:
.evals/evals-2/kb.json(shared memory) - Logs:
scripts/deploy/logs/*.log - Reports:
.evals/evals-2/*.json
Deployment Safety
Before git commit:
./continuous-frontend-nextjs16-qa-compliance-eval-system.sh once
# Only commit if you see:
# β
Deployable: YESTroubleshooting
"Frontend not accessible"
cd clients/frontend
npm run dev"Playwright not found"
pip3 install playwright
python3 -m playwright install chromium"Too many errors"
# Run fix cycle
./continuous-frontend-improve-from-eval-agent-reports.sh onceNext Steps
- β Run for a few hours
- π Review kb.json learnings
- π― Adjust thresholds in scripts
- π§ Add custom checks
- π Integrate with CI/CD
See the Agent Framework Generalization Guide for full documentation.