Devops Infra

• Direct process execution (no PM2 for local)
• Dashboard API: node server.js (port 3001)
• Frontend: npm run dev (port 5173)
• OMP daemon via dashboard API

Config and Secrets Management

• Local Secrets: dashboard/.env (never in git)
• CI Secrets: GitHub Secrets (DATABASE_URL, BRAPI_API_KEY)
• Validation: Check required vars before running

Observability (Local)

• Health Checks: scripts/local-health-check.sh (planned)
  • API health, OMP status
  • System resources: CPU%, Memory%, Disk%
• Logs: Dashboard server logs, campaign output
• Monitoring: .github/workflows/monitoring.yml (daily data checks)

Reliability (Local)

• Cleanup: scripts/auto_cleanup.sh for disk management
• Health Checks: Cron-compatible for monitoring
• Graceful Shutdown: Signal handlers

Local Resource Management

• Monitor with htop, df -h, free -h
• Cleanup with scripts/cleanup_old_runs.sh
• Configure workers based on CPU cores

CI/CD

• CI: .github/workflows/ci.yml (tests, benchmarks, calendar)
• Monitoring: .github/workflows/monitoring.yml (daily data checks)

VPS Infrastructure - DEFERRED

VPS deployment is DEFERRED. Scripts in scripts/vps/ are for historical reference. See docs/ops/local_only_policy.md.

When to Use

INVOKE this skill when:

• Local build failed or service not starting
• Memory or disk issues on local workstation
• DB connections saturated
• Setup or troubleshoot CI/CD
• Secrets exposed or environment drift
• Health checks failing
• Cleanup needed for disk space

DO NOT use this skill when:

• Optimizing application code (use /quant-engineer)
• Investigating data quality (use /data-engineer)
• Validating strategy performance (use /risk-analyst)
• Designing strategy logic (use /quant-researcher)

Operating Rules

Hard Constraints

1. Never run without traceable git sha

   • Use git log -1 --format="%H" before major operations

2. Never modify env without logging change

   • Document what changed, when, why
   • Keep ops change log

3. Never commit secrets to git

   • Use env vars and GitHub Secrets
   • Template files only in repo

4. Never update schema without migration + rollback plan

   • Coordinate with data-engineer
   • Test rollback before applying

5. Never ignore disk full / memory pressure alerts

   • Monitor with df -h and free -m
   • Clean artifacts: scripts/cleanup_old_runs.sh

6. Never operate without backup/restore plan

   • Neon PostgreSQL has automatic backups
   • Local artifacts backed up before deletion

7. Never change configs without local validation first

   • Test configs before running campaigns

Repo Anchors

Local Scripts

CI/CD Workflows

VPS Scripts (DEFERRED)

Quick Reference

Local Commands

# Start dashboard API
cd dashboard && node server.js

# Start frontend (dev)
cd dashboard && npm run dev

# Build frontend
cd dashboard && npm run build

# Check API health
curl http://localhost:3001/api/health

# Cleanup old runs
./scripts/auto_cleanup.sh --runs --days 3

Monitoring Commands

# System resources
htop
free -h
df -h

# Disk usage by directory
du -sh output/scg/
du -sh target/
du -sh artifacts/