Usage

Running migrations & anything on the preview

D1 migrations are applied automatically before deploy (run-migrations: true). For everything else, use the command hooks — they run with the resolved preview environment injected as env vars:

Env var	Meaning
PREVIEW_WORKER	preview worker name
PREVIEW_DB	preview D1 name
PREVIEW_DB_ID	preview D1 id (deploy hooks)
PREVIEW_URL	preview URL (deploy hooks)
PREVIEW_CONFIG	path to the rendered wrangler.preview.jsonc (deploy hooks)
WRANGLER_COMMAND	the configured wrangler invocation

      - uses: BitByBit-B3/cf-pr-preview-action@v1
        with:
          worker-prefix: myapp-preview
          post-deploy-command: |
            $WRANGLER_COMMAND d1 execute "$PREVIEW_DB" --config "$PREVIEW_CONFIG" --remote --file ./seed/preview.sql
            $WRANGLER_COMMAND d1 execute "$PREVIEW_DB" --config "$PREVIEW_CONFIG" --remote --command "INSERT INTO flags (k,v) VALUES ('preview',1)"
        env:
          CLOUDFLARE_API_TOKEN: ${{ secrets.CLOUDFLARE_API_TOKEN }}
          CLOUDFLARE_ACCOUNT_ID: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}

There are three hooks: pre-deploy-command, post-deploy-command, and pre-teardown-command.

Using the outputs

Prefer your own steps? Skip the hooks and consume the outputs:

      - uses: BitByBit-B3/cf-pr-preview-action@v1
        id: preview
        with:
          worker-prefix: myapp-preview
        env:
          CLOUDFLARE_API_TOKEN: ${{ secrets.CLOUDFLARE_API_TOKEN }}
          CLOUDFLARE_ACCOUNT_ID: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}
      - run: echo "Preview at ${{ steps.preview.outputs.url }}"

Package managers

wrangler-command defaults to bunx wrangler. For npm/pnpm:

        with:
          worker-prefix: myapp-preview
          wrangler-command: npx wrangler   # or: pnpm dlx wrangler

Choosing the config env

If your wrangler.jsonc uses named environments, config-env (default main) selects which env.<name> block to base the preview on. Global (top-level) vars are merged in too; the env block wins on key collisions.