KGC Dev System

Error Detection → Fix Pipeline

A single system for building, shipping, and keeping apps healthy. When something breaks in production, you know where to look — and the pipeline from error to fix is already defined.


The full stack

Six tools that cover detection, alerts, tracking, and fixes:

ToolJob
SentryDetects errors in live apps automatically
SlackReceives alerts from Sentry + Linear
LinearTracks all bugs and tasks
Cursor Cloud AgentAI that attempts to fix the error
UptimeRobotMonitors if sites are up or down
Vercel AnalyticsFrontend traffic and usage stats

Error pipeline

From production error to closed ticket:

1. Error occurs in live app
         ↓
2. Sentry detects it automatically
         ↓
3. Sentry → Slack (Alert #1: what broke + where)
   Sentry → Linear (issue created and logged)
         ↓
4. Linear → Slack (Alert #2: issue confirmed tracked)
         ↓
5. Open Sentry → find the error → read the Root Cause
         ↓
6. Click "Send to Cursor Cloud Agent"
         ↓
7. Cursor Agent analyzes the error and attempts the fix
         ↓
8. Click "Open in Cursor" → review the fix
         ↓
9. Test locally → confirm fix works
         ↓
10. Push to GitHub → auto deploys to Vercel / Railway
         ↓
11. Sentry confirms error stops occurring in production
         ↓
12. Close the Linear issue

Build + deploy pipeline

From idea to live URL:

1. ChatGPT        → Research
2. Claude         → Architecture, concepts, brand
3. Cursor         → Full build and debug
4. Local          → Test everything before shipping
5. Gitea          → Local git commit
6. GitHub         → Cloud git push
7. Railway        → Backend auto deploys
   Vercel         → Frontend auto deploys
8. UptimeRobot    → Auto-monitors the new live URL
9. AWS / Supabase → Storage if needed
10. Sentry        → Watching for errors in production

Platform rules

One place per job — no guessing where things live:

WhatWhere
Frontend (React, Next.js, static sites)Vercel
Backend (APIs, servers, workers)Railway
Database (PostgreSQL)Railway
File storageAWS S3
Error trackingSentry
Uptime monitoringUptimeRobot
Bug + task trackingLinear
Team alertsSlack

Account structure

  • Vercel — One account per billing entity (your projects vs client projects).
  • Railway — One account per billing entity.
  • Credentials — All in Bitwarden; never in plain text.
  • Notes — One Obsidian note per project (see PROJECT-TEMPLATE.md).

Pre-demo checklist

Before showing any project to anyone:

  • Live URL — Open it; confirm it loads.
  • Linear — Check the backlog; know what’s broken before they see it.
  • User flow — Test the main path end to end.
  • UptimeRobot — No recent downtime.
  • Data — Demo data or test accounts are ready.

The one habit that holds it all together

When you find a bug — open Linear and write one line before you close your laptop.
Don’t fix it now. Just log it. Memory is not a system.


KGC Dev System — March 2026