Build your first tool

Publish a private Node tool from the Scrapy My tools → Develop New flow. Scrapy builds your source into a Cloudflare Container image; your code talks to the runtime API with a run-scoped token.

What to upload

Required	Optional
package.json with scripts.start and @scrapyio/sdk ^0.2.0	input_schema.json
src/main.js (entry)	README.md
package-lock.json (for npm ci)	—

Do not include a Dockerfile. Scrapy generates one from the platform template catalog (node-basic for plain JS).

Minimal entry

import { Tool, log } from '@scrapyio/sdk';

await Tool.init();
const input = (await Tool.getInput()) ?? {};
const message = typeof input.message === 'string' ? input.message : 'hello-from-starter';
await Tool.pushData([{ message }]);
await Tool.exit({ statusMessage: 'ok' });

Input → output example

{ "message": "hello-from-demo" }

Run results include an item with that same message string.

Develop New (template)

1. Open My tools → Develop New.
2. Choose Web scraper → Container → a Node template → preview → Use this template.
3. Click Save. Scrapy creates a private tool you own and seeds the Node starter.
4. On the tool page, confirm Source (or edit in the Web IDE and Save source).
5. Open Last build → Build. Wait until status is Ready.
6. Open Input / Last run → Test run with JSON input. Check results and logs.

Import from Git

1. On Develop New, click Import from Git.
2. Scrapy creates an empty private tool and opens the GitHub import panel.
3. Connect the Scrapy GitHub App if needed, pick repo / branch / subfolder, import.
4. Build → Ready → Test run (same as above).

Download starter

On the Source panel, use Download starter template (scrapy-node-starter.zip), or GET /api/publisher/tools/starter.zip while signed in.

Limits (v1)

• Node only (no Python publisher builds yet).
• Environment variables / secrets for tools are coming soon (K25). Containers never receive DATABASE_URL or account secrets.