Deploying on Railway

Railway is the lowest-friction path. Click the deploy button, paste your env vars, and you're live.

One-click deploy

The Mira README has a Deploy on Railway button. It clones the railway.toml config:

[build]
builder = "nixpacks"

[build.dockerImage]
image = "ghcr.io/miracodeai/mira:latest"

[deploy]
healthcheckPath = "/health"

Railway pulls the image, exposes a public URL, and points its health check at /health.

Set env vars

In the Railway service's Variables tab, set:

MIRA_GITHUB_APP_ID=123456
MIRA_GITHUB_PRIVATE_KEY=<paste full PEM, including BEGIN/END lines>
MIRA_WEBHOOK_SECRET=<32+ bytes random>
OPENROUTER_API_KEY=sk-or-v1-...
MIRA_MODEL=anthropic/claude-sonnet-4-6
ADMIN_PASSWORD=<change from default>
MIRA_DASHBOARD_URL=<paste the public URL Railway gives you>

Pasting a multi-line PEM into Railway's variable editor works; keep the newlines. Don't escape them as \n.

Add a database

Click + New → Database → PostgreSQL in the same project. Railway will inject a DATABASE_URL reference variable; bind it to your Mira service:

DATABASE_URL=${{Postgres.DATABASE_URL}}

Mira picks this up automatically. Append ?sslmode=require for production.

If you'd rather use SQLite, skip the Postgres step and add a Railway volume mounted at /data/indexes instead.

Point the GitHub App at Railway

Once deployed, copy the public URL Railway gives the service (something like https://mira-production.up.railway.app) and set the GitHub App's Webhook URL to:

https://mira-production.up.railway.app/github/webhook

Reinstall (or trigger a webhook redelivery from the GitHub App's Advanced settings) to confirm Mira is receiving events.

Verifying

curl https://mira-production.up.railway.app/health
# → 200 OK

Open the public URL in a browser and log in with admin / <ADMIN_PASSWORD>.

Resource sizing

Mira runs comfortably on Railway's smallest plan for small/medium repos. If you're indexing very large repos, bump the memory allocation; indexing is where Mira spikes RAM.