Imported from jeremylongshore/tons-of-skills-marketplace (skills/.curated/supabase-ci-integration/SKILL.md). Install upstream with npx skills add jeremylongshore/tons-of-skills-marketplace --skill supabase-ci-integration. Copyright stays with the author (MIT).

Supabase CI Integration

Overview

Build GitHub Actions workflows that automate the full Supabase lifecycle: link projects in CI, push migrations on merge, deploy Edge Functions, generate TypeScript types, run tests against a local Supabase instance, and create preview branches for pull requests. Every database change gets validated before it reaches production.

The pull-request CI pipeline runs these stages, all detailed in ci-workflows.md:

1. Start a local Supabase instance (unused services disabled for speed).
2. Apply every migration from scratch with npx supabase db reset.
3. Regenerate TypeScript types and fail the build on drift from the committed version.
4. Run pgTAP database tests and the application test suite against the local instance.
5. Stop Supabase (always, even on failure).

Prerequisites

• GitHub repository with Actions enabled
• Supabase project created at supabase.com/dashboard
• Supabase CLI initialized locally (npx supabase init)
• Node.js 18+ in your project
• @supabase/supabase-js installed:

npm install @supabase/supabase-js

Instructions

Step 1: Configure GitHub Secrets and Link in CI

Store credentials as GitHub repository secrets. The CI pipeline uses these to authenticate with your Supabase project without exposing tokens in code.

# Set secrets via GitHub CLI
gh secret set SUPABASE_ACCESS_TOKEN --body "<your-access-token>"
gh secret set SUPABASE_DB_PASSWORD --body "<your-database-password>"
gh secret set SUPABASE_PROJECT_REF --body "<your-project-ref>"

Generate your access token at supabase.com/dashboard/account/tokens. Find your project ref in Project Settings > General.

Link the project in any CI job that needs remote access:

- name: Install Supabase CLI
  uses: supabase/setup-cli@v1
  with:
    version: latest

- name: Link Supabase project
  run: npx supabase link --project-ref ${{ secrets.SUPABASE_PROJECT_REF }}
  env:
    SUPABASE_ACCESS_TOKEN: ${{ secrets.SUPABASE_ACCESS_TOKEN }}

Step 2: CI Workflow — Test, Validate Migrations, and Generate Types

Add .github/workflows/supabase-ci.yml that runs on every pull request. It starts a local Supabase instance, runs db reset to apply all migrations from scratch, regenerates types and fails on drift (git diff --exit-code), then runs pgTAP and application tests. The default local dev keys are safe to commit — they only work against the local instance. Copy the complete workflow from ci-workflows.md.

Step 3: Deploy Migrations and Edge Functions on Merge

Add .github/workflows/supabase-deploy.yml, scoped with paths: to supabase/migrations/** and supabase/functions/** so it only runs when database or function code changes on main. It links the remote project, runs npx supabase db push, deploys Edge Functions, and regenerates types from the production schema. Full workflow in ci-workflows.md.

Preview Branches

Create isolated Supabase environments per pull request with npx supabase branches create, so reviewers test against real infrastructure with migrations applied. Preview branches require a Supabase Pro plan and incur compute costs while running. Workflow snippet in ci-workflows.md.

Testing in CI

Two test layers run inside the CI workflow:

• pgTAP database tests in supabase/tests/ validate that RLS is enabled on public tables and that policies behave correctly. Run locally with npx supabase test db.
• Application tests use createClient from @supabase/supabase-js pointed at http://127.0.0.1:54321 with the local anon key.

Both patterns — the pgTAP SQL and the TypeScript client setup — are in testing-patterns.md.

Output

After implementing these workflows:

• Pull requests run tests against a fresh local Supabase instance with all migrations applied
• TypeScript type drift is detected automatically — stale types block the PR
• Database migrations deploy to production only on merge to main
• Edge Functions deploy alongside migration changes
• pgTAP tests validate RLS policies and schema constraints in CI
• Preview branches provide isolated environments for PR review (Pro plan)
• GitHub secrets keep SUPABASE_ACCESS_TOKEN and SUPABASE_DB_PASSWORD out of code

Error Handling

Examples

Minimal CI for a new project — just migration validation and type checking:

name: Supabase CI
on: [pull_request]
jobs:
  validate:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: supabase/setup-cli@v1
        with: { version: latest }
      - run: npx supabase start -x realtime,storage-api,imgproxy,inbucket,edge-runtime
      - run: npx supabase db reset
      - run: npx supabase gen types typescript --local > /tmp/types.ts && diff src/types/database.types.ts /tmp/types.ts
      - if: always()
        run: npx supabase stop

For the full CI + deploy + preview workflows and an Edge Function deploy-with-verification snippet, see ci-workflows.md.

Resources

• Supabase CLI in CI/CD — official setup guide
• Database Testing with pgTAP — writing and running database tests
• Managing Environments — staging, production, preview branches
• Edge Functions Deployment — deploying Deno functions
• Type Generation — keeping TypeScript types in sync
• Branching & Preview — per-PR database environments
• @supabase/supabase-js Reference — client SDK documentation

Next Steps

For deploying Supabase-backed applications to hosting platforms, see supabase-deploy-integration. For configuring RLS policies, see supabase-rls-policies.