Builds & the deploy pipeline

Every deploy — whether from chat, git, or upload — flows through the same build pipeline. Cantila figures out how to build your code, attaches your project's services, and ships an image.

Stack detection

Cantila detects your stack automatically with Nixpacks: it inspects the files in your project and picks the right runtime and build steps. If your project includes a Dockerfile, Cantila honors it instead and builds the container exactly as defined — full control when you need it.

Supported runtimes

Runtime	Detected from / use for
`static`	Static sites and pre-built front-ends
`node`	Node.js apps (Next.js, Express, and similar)
`python`	Python apps (FastAPI, Flask, Django)
`php`	PHP applications
`go`	Go services
`ruby`	Ruby apps
`docker`	Any container — your `Dockerfile` is the source

Set the runtime explicitly when creating a project:

cantila create my-app --runtime python --region fsn1

The 8-step deploy pipeline

Each deploy runs the same ordered pipeline. Before the build runs, Cantila auto-wires the project's managed database, mailbox, and SMS number and injects roughly eight connection secrets as environment variables, so your code can reach its services without any manual configuration.

Receive the source (repo push, upload, or chat-assembled files)
Auto-wire the project's DB, mailbox, and SMS number
Inject the connection secrets as env vars
Detect the stack with Nixpacks (or honor a Dockerfile)
Build the image
Publish the image
Roll out the new deployment
Promote it to live and expose the URL

You watch this stream step by step — see streaming build logs.

Because the connection secrets are injected automatically, you don't set DATABASE_URL and friends yourself. Add your own variables on top — see env & secrets.

Build configuration

For git projects, Cantila reads a cantila.json at the repo root when present, then falls back to package.json build scripts for Node, then to the detected stack's defaults. See push-to-deploy for an example cantila.json.