Forest Admin - API reference

Forest agent for Node.js: your data, workflows and AI agents, running on your infrastructure.

Forest is the back office for operations teams in regulated companies. Ops teams review, decide and act on your data from one interface, or let AI agents do it through MCP, within the roles, permissions and audit logs you define.

This repository is the Node.js agent: the part of Forest that runs on your servers, connects to your databases and services, and exposes them to the Forest interface and to AI agents. Your business data never goes through Forest.

Note

Forest v2. This is the current Node.js SDK. Using forest-express-sequelize or forest-express-mongoose? You are on v1: see the migration guide.

Quick start

npm install @forestadmin/agent @forestadmin/datasource-sql

import { createAgent } from '@forestadmin/agent';
import { createSqlDataSource } from '@forestadmin/datasource-sql';

const agent = createAgent({
  authSecret: process.env.FOREST_AUTH_SECRET,
  envSecret: process.env.FOREST_ENV_SECRET,
  isProduction: process.env.NODE_ENV === 'production',
});

agent.addDataSource(createSqlDataSource(process.env.DATABASE_URL));

await agent.start();

Required environment variables: FOREST_ENV_SECRET, FOREST_AUTH_SECRET, DATABASE_URL. Full guide: docs.forest.app/get-started/quickstart

Requirements


Node.js	Developed and tested on Node.js 24 (see `.nvmrc`)
Framework	None required: standalone server, or mounted on Express, Koa, Fastify or NestJS

What the agent gives you


Data	Connect your databases and services through data sources: SQL, Sequelize, Mongoose, MongoDB, Zendesk, replicas, RPC.
Customization	Computed fields, actions, segments, hooks and validations in TypeScript, with autocompletion generated from your schema.
Workflows	Runs your Forest business processes on your side (`@forestadmin/workflow-executor`).
MCP server	Exposes your collections and actions to AI agents through the Model Context Protocol (`@forestadmin/mcp-server`).
Permissions and audit	Roles, scopes and activity logs defined in Forest, enforced on every request, from the UI or from an AI agent.

How it works

The Node.js agent runs on your infrastructure. Forest calls it over HTTPS; the agent queries your databases and services, and exposes an MCP server to AI agents.

The agent runs on your infrastructure. The Forest interface calls your agent, and your agent queries your sources: Forest does not store your business data. You control what is exposed, to whom, and the logic applied to it.

Documentation


Get started	https://docs.forest.app/get-started/quickstart
Connect a data source	https://docs.forest.app/get-started/connect/data-sources/overview
Node.js API reference	https://docs.forest.app/reference/agent-api/nodejs
MCP server	https://docs.forest.app/product/embed/mcp-server
Migrating from v1	https://docs.forest.app/guides/migration/from-v1/overview

Packages in this monorepo

@forestadmin/agent · datasource-sql · datasource-sequelize · datasource-mongoose · datasource-mongo · datasource-zendesk · datasource-replica · datasource-customizer · mcp-server · workflow-executor · agent-testing · plugin-aws-s3 · plugin-flattener · plugin-export-advanced · forest-cloud

Experimental plugins and data sources: forestadmin-experimental.

Contributing

Bug reports and pull requests are welcome. See CONTRIBUTING.md for setup, tests and commit conventions.

yarn install && yarn build && yarn test

License

Apache 2.0. See LICENSE.