Self Host on Vercel

Vercel runs container images on Vercel Functions with autoscaling and Active CPU pricing.

Add a Dockerfile.vercel

FROM evidencedev/serve:latest
COPY --chown=evidence:evidence . /project

Modify and commit connection.yaml

The project must use a direct connector, with secrets replaced by ${VAR} environment references:

type: snowflake
account: xy12345.us-east-1
user: ${SNOWFLAKE_USER}
private_key: ${SNOWFLAKE_PRIVATE_KEY}
warehouse: COMPUTE_WH
database: ANALYTICS

connection.yaml is gitignored by default. Once its secrets are ${VAR} placeholders, remove it from .gitignore and commit it — the container reads it at runtime.

Deploy

Using the Vercel UI
Using the Vercel CLI

Go to vercel.com/new and import the project’s GitHub repo.In the Environment Variables section of the import screen, add:

EVIDENCE_BASIC_USER — HTTP Basic Auth username
EVIDENCE_BASIC_PASSWORD — HTTP Basic Auth password
every ${VAR} referenced in connection.yaml, here SNOWFLAKE_USER and SNOWFLAKE_PRIVATE_KEY

Click Deploy.

Create the project:

vercel link

Accept the defaults — “No framework detected” is expected.Set the environment variables:

vercel env add SNOWFLAKE_USER production
vercel env add SNOWFLAKE_PRIVATE_KEY production
vercel env add EVIDENCE_BASIC_USER production
vercel env add EVIDENCE_BASIC_PASSWORD production

Repeat with preview to make preview deployments work too. Then deploy:

vercel deploy --prod

For automatic deploys on every push, connect the repo:

vercel git connect

Pushes to the default branch then deploy to production; other branches and PRs get preview URLs.

No PORT variable is needed — Vercel provides one and the image binds to it automatically.

Notes

Container Images are in Beta on Vercel.
Functions scale to zero when idle; the first request after idle incurs a cold start.

⌘I

Get Started

Core Concepts

Features

MCP

CLI

Self Host

Warehouses

Evidence Warehouse

Components

Notes

​Notes

Notes