After living through Texas grid scares, I wanted a single page that answers the only questions most people actually have: is the grid under stress right now, where is the power coming from, and are prices spiking? Not a wall of engineering telemetry, just a plain-English read in about 30 seconds.
That's ERCOT Grid Status: a free dashboard for non-technical Texans, hosted entirely on GitHub.
The idea
ERCOT publishes a lot of public data, but it's aimed at people who already understand the grid. The gap I wanted to close was translation: taking demand, fuel mix, pricing, and outage data and turning it into three status tiles anyone can read at a glance.
What it shows
- Status tiles: grid stress (Normal / Watch / Stressed), renewable-energy percentage, and a price indicator
- 7-day charts: demand trends, fuel-mix breakdown, real-time pricing, and outage patterns
- A transparent stress classification with the logic spelled out, so the "Stressed" label is never a black box
The architecture
The interesting part is that there's no server anywhere. The whole thing runs on GitHub's free tier:
- A GitHub Actions job runs every 15 minutes, authenticates to the ERCOT Public Data API, and pulls the latest readings
- TypeScript normalizes the raw feed and computes the grid-stress classification
- The results are written as rolling 7-day JSON files committed back into the repo, which gives me free history and a fully auditable data trail
- A static site on GitHub Pages reads those JSON files and renders the tiles and charts
Why commit the data to git
Storing snapshots in version control instead of a database was a deliberate choice. It means the data has a transparent history anyone can inspect, there's nothing to provision or pay for, and the entire pipeline (fetch, transform, publish) is reproducible from the repo alone.
Keeping it secure and free
The ERCOT credentials live in GitHub Secrets and are only ever used inside the Actions runner, so they never reach the browser. Because everything downstream is static JSON and a static page, the dashboard costs nothing to run and has no attack surface tied to user data.
The design constraint
The hardest part wasn't the plumbing. It was deciding what "Stressed" means. I kept the classification conservative and clearly explained, because a grid-status tool is only useful if people trust the label. Getting that translation right is the actual product.
Key Takeaways
GitHub Actions plus GitHub Pages can run a real, self-updating data product with no servers and no hosting bill.
Committing rolling JSON snapshots to the repo gives you free history and a fully transparent data trail.
Translating raw grid telemetry into a conservative, clearly-explained status is a product decision, not just a data one.
Keeping API credentials in GitHub Secrets means they never touch the browser on a fully static site.
