zero-cost-ops

Production monitoring for solo founders. $0/month. No excuses.

CTO Cockpit — traffic-light semaphores, incident log, morning sync

System Dashboard — live metadata from Supabase via n8n

What is this?

A complete CTO-grade operations dashboard and automation pipeline that runs entirely on free tiers. Built by a solo founder who got tired of paying for monitoring tools that did less than what she could build in a weekend.

Monitors n8n workflows, Slack channels, GitHub repos, Cloudflare, LinkedIn, and Meta — with traffic-light semaphores, incident tracking, and a 60-second morning sync. Ingests content automatically via webhook, classifies it with Claude AI, cross-links related entries, and builds a searchable metadata layer in Supabase. Ships as a PWA you can pin to your phone and open at 7am to know exactly what’s broken (and what isn’t).

This is the real system. Not a demo. Open-sourced so other solo founders don’t have to build it from scratch.

Architecture

┌─────────────────────────────────────────────────────────────────┐
│                        DATA SOURCES                             │
│  Slack    GitHub    Cloudflare    LinkedIn    Meta    Webhooks   │
└────────────────────────────┬────────────────────────────────────┘
                             │  HTTP POST to webhook
                             ▼
┌─────────────────────────────────────────────────────────────────┐
│                    n8n (Render free tier)                       │
│                                                                  │
│  WF-01: Ingest → Claude classify → Supabase insert → trigger   │
│  WF-04: Timeline → temporal cluster → narrative arc → update    │
└────────────────────────────┬────────────────────────────────────┘
                             │  REST API (anon key)
                             ▼
┌─────────────────────────────────────────────────────────────────┐
│                   Supabase (free tier)                          │
│                                                                  │
│  PostgreSQL 15 · RLS · Real-time · 5 tables · GIN indexes       │
└────────────────────────────┬────────────────────────────────────┘
                             │  fetch() from browser
                             ▼
┌─────────────────────────────────────────────────────────────────┐
│               GitHub Pages (free, custom domain)                │
│                                                                  │
│  dashboard/cto-cockpit.html   →   Your phone (PWA)             │
│  dashboard/system-dashboard.html  →  Status overview            │
└─────────────────────────────────────────────────────────────────┘

The Stack (and what it costs)

Service	What it does	Free tier limit	Cost
n8n on Render	Automation workflows, webhook ingestion, Claude AI calls	750 hours/month (always on with keep-alive)	$0
Supabase	PostgreSQL database, auth, real-time, RLS	500MB storage, 50K API calls/month	$0
GitHub Pages	Static dashboard hosting, custom domain, HTTPS	100GB bandwidth/month, unlimited sites	$0
Cloudflare	DNS, DDoS protection, analytics	Unlimited DNS, 100K requests/day	$0
UptimeRobot	External uptime monitoring, 5-min checks	50 monitors	$0
Claude API	Content classification (optional)	Pay-as-you-go, ~$0.01/classification	$0*

Total: $0/month (Claude API optional — disable the classify step to stay fully free)

Features

🟢 Traffic-light semaphores — green/yellow/red status for every service at a glance
⚡ Auto-ingestion pipeline — webhook receives content, Claude classifies it, Supabase stores it
🤖 AI classification — Claude API extracts tags, entities, narrative role, timeline group
🔗 Cross-linking engine — related entries automatically linked by semantic similarity
📅 Narrative timeline — content clustered by time period and narrative arc
📱 PWA mobile-first — install to home screen, offline-capable service worker
🌑 Dark cockpit theme — monospace, #0a0a0a background, easy on eyes at 7am
☁️ Zero-infrastructure — no servers to manage, no Docker, no DevOps

Quick Start

You’ll have a working dashboard in under 30 minutes.

1. Fork this repo

# Fork on GitHub, then:
git clone https://github.com/YOUR_USERNAME/zero-cost-ops.git
cd zero-cost-ops

2. Set up Supabase

Create a free account at supabase.com
Create a new project (choose EU region for GDPR if relevant)
Go to SQL Editor and run schema/supabase-schema.sql
Copy your project URL and anon key from Settings > API

3. Deploy n8n on Render

Create a free account at render.com
New > Web Service > Docker
Use image: n8nio/n8n:latest

Set environment variables:

N8N_BASIC_AUTH_ACTIVE=true
N8N_BASIC_AUTH_USER=admin
N8N_BASIC_AUTH_PASSWORD=your-secure-password
SUPABASE_URL=https://YOUR-PROJECT.supabase.co
SUPABASE_SERVICE_ROLE_KEY=your-service-role-key

Note your Render URL: https://your-n8n.onrender.com

4. Import workflows

In n8n, go to Workflows > Import from File
Import workflows/WF-01-ingestion-pipeline.json
Import workflows/WF-04-timeline-engine.json
Update Supabase credentials in each workflow
Activate both workflows

5. Configure the dashboard

Edit dashboard/system-dashboard.html — find the API constant and update:

const API = 'https://YOUR-N8N-INSTANCE.onrender.com/webhook/system';

Edit dashboard/cto-cockpit.html — update the CONFIG object with your services.

6. Enable GitHub Pages

Go to your repo Settings > Pages
Source: Deploy from a branch > main > / (root)
(Optional) Add custom domain via Cloudflare

7. Add keep-alive for Render free tier

Render free instances sleep after 15 minutes of inactivity. Add a UptimeRobot monitor to ping your n8n health endpoint every 5 minutes:

https://YOUR-N8N-INSTANCE.onrender.com/healthz

That’s it. Your CTO cockpit is live.

Workflows Included

WF-01: Ingestion Pipeline

What it does: Receives content via webhook → filters noise → classifies with Claude AI → stores in Supabase → triggers cross-linking.

Webhook → Extract Content → Filter → Claude Classify → Parse → Supabase Insert → Trigger WF-02 → Respond

Inputs: Any POST to /webhook/wf01-ingest with { text, source, user_id }
Output: Classified entry in Supabase metadata table with tags, entities, narrative role, and timeline group
Claude model: claude-sonnet-4-20250514

WF-04: Timeline Engine

What it does: Scans all entries in Supabase → clusters by time period → detects narrative arc patterns → updates timeline_group field in bulk.

Webhook → Fetch All → Temporal Clustering → Narrative Arc Detection → Batch Update → Summary

Run: Trigger manually or on a schedule (weekly recommended)
Output: All entries tagged with 1974-1990 | 2019-2023 | 2024-2026 + narrative role analysis

See workflows/README.md for detailed documentation.

Customization

This system is designed to be adapted, not just deployed. You can:

Change monitored services — edit the CONFIG object in cto-cockpit.html
Add new workflows — follow the same pattern as WF-01 (webhook → process → Supabase)
Modify the schema — add columns to metadata or create new tables
Replace Claude — swap the HTTP request node for any other LLM API
Add authentication — Supabase Auth is already configured in the schema

See docs/CUSTOMIZATION.md for a full guide.

Live Demo

Main app: neyen.thespiralwithin.ai
System dashboard: neyen.thespiralwithin.ai/system.html

Contributing

Issues, PRs, and feedback welcome. See CONTRIBUTING.md.

Built by

Ana Ballesteros Benavent
CTO, NEYEN / The Spiral Within SLU
Valencia, Spain

thespiralwithin.ai · GitHub: ProyectoAna

Built this because I needed it. Sharing it because you might too.

License

MIT — see LICENSE. Use it, fork it, ship it.

This site is open source. Improve this page.