Notion Developer Platform launched May 13, 2026. Complete guide to Workers, External Agents API (Claude Code, Codex, Cursor), Database Sync, pricing, and setup.
Notion launched its Developer Platform on May 13, 2026, turning Notion from a productivity workspace into a programmable, agent-ready operating system for teams. The release introduces three capabilities in one: Workers, a serverless hosted runtime for custom JavaScript and TypeScript code running inside Notion’s own infrastructure; an External Agents API, which brings Claude Code, Cursor, OpenAI Codex, and Decagon directly into your workspace as first-class participants; and Database Sync, which pulls any REST or GraphQL API into a live Notion database without external servers. If you build with Notion and work with AI agents, this changes your architecture options. You can now run custom logic, connect agentic AI tools, and mirror external data entirely within Notion’s platform. This guide covers every major component: how Workers are triggered, how the External Agents API authorization model works, what Database Sync supports at launch, the pricing timeline, and the real-world implications for teams building AI-powered workflows.
What the Notion Developer Platform Actually Is
For the past decade, extending Notion in any meaningful way required either third-party automation tools like Zapier or Make, or the Notion API backed by self-hosted infrastructure you maintain. The Developer Platform eliminates the hosting dependency. Workers run inside Notion’s own infrastructure. External agents gain workspace access through a standardized protocol. Database Sync replaces the manual export-transform-import cycle with live, Worker-powered connections that stay current automatically.
The platform has three separate layers that work independently or together:
- Notion Workers: A hosted JavaScript/TypeScript runtime with full read/write access to your workspace
- External Agent API: A protocol that lets external AI agents show up in your workspace as first-class participants you can interact with directly
- Database Sync: A Workers-powered connector that keeps Notion databases in sync with external systems that have an API
All three are available today for Business plan workspaces and above. Free and Plus plans are excluded at launch.
Notion Workers: Serverless Logic Inside Your Workspace
Workers are the core primitive. A Worker is a JavaScript or TypeScript function that runs inside Notion’s own hosted sandbox — no AWS Lambda, Vercel Functions, or self-managed server required. Workers have full access to your Notion workspace through a built-in SDK and can call external APIs as needed. They run in response to triggers, not on a persistent schedule, which keeps execution costs predictable.
Install the Notion CLI to get started:
curl -fsSL https://ntn.dev | bashOnce installed, the CLI covers the full development lifecycle:
# Authenticate with your workspace
notion login
# Create a new Worker project
notion worker init my-worker
# Deploy to Notion's hosted infrastructure
notion worker deploy
# Test a worker manually
notion worker invoke my-worker --payload '{"page_id": "abc123"}'Three Trigger Types
Workers can be triggered in three distinct ways, and the trigger model determines which integration pattern to use.
Agent Tools: Attach a Worker to a Custom Notion Agent as a callable tool. When the agent determines the tool is relevant to a task, it invokes the Worker automatically. This is the most powerful pattern for AI-driven automation inside Notion: the agent reasons about when to act, the Worker executes custom logic, and Notion records the result. If you are building an AI assistant that lives in Notion and needs to interact with external systems, Agent Tools is the trigger type to use.
Webhooks (now bidirectional): Notion webhooks were previously outbound-only — Notion could push events to external systems when pages or database entries changed, but external systems could not push events back into Notion without a separately hosted endpoint. Workers fix this. Each deployed Worker has a webhook URL. Any external system can POST to that URL, triggering the Worker to run logic and take action in Notion. A CRM can push a new deal into a Notion pipeline database. A CI system can create a Notion incident page when a build fails. The integration model shifts from “pull data out of Notion” to “push events into Notion.”
Database Sync: Workers can be designated as sync providers for a Notion database. Notion calls the Worker on a configured schedule or on demand, the Worker fetches data from an external API, and Notion updates the database accordingly. This powers the live external data sync use case described in its own section below.
A Real Worker: Slack Bug Reports to Notion
A practical Worker that listens for Slack messages tagged with #bug and creates a prioritized Notion database entry:
import { Client } from '@notionhq/client'
export default async function handler(req) {
const notion = new Client({ auth: process.env.NOTION_TOKEN })
const { text, user, channel } = req.body
if (!text.includes('#bug')) return new Response('skipped', { status: 200 })
const priority = text.toLowerCase().includes('urgent') ? 'High' : 'Medium'
await notion.pages.create({
parent: { database_id: process.env.BUG_DATABASE_ID },
properties: {
Title: { title: [{ text: { content: text.slice(0, 100) } }] },
Priority: { select: { name: priority } },
Reporter: { rich_text: [{ text: { content: user } }] },
Channel: { rich_text: [{ text: { content: channel } }] },
},
})
return new Response('created', { status: 200 })
}Deploy this Worker, configure its webhook URL in Slack’s event subscriptions, and every #bug message in any Slack channel creates a Notion entry automatically — no external server required. The same pattern applies to GitHub issue creation, Stripe payment alerts, Zendesk ticket escalations, or any system that can POST to a URL.
Comments · 0
No comments yet. Be the first to share your thoughts.