Browser Extension

Slack

**MCP** (Model Context Protocol) is an open standard for exposing tools and data sources to AI agents. River is fully MCP-compatible, which means anything that runs as an MCP server can become a set of tools the AI can use.

This is a fast path to extending River's capabilities — far faster than waiting for an official integration.

## What MCP unlocks

There's a growing ecosystem of MCP servers for popular services and protocols:

- Databases (Postgres, MongoDB, Snowflake, BigQuery)
- Dev tools (Jira, Sentry, PagerDuty, Datadog)
- Niche SaaS that doesn't have an official River integration yet
- Internal company tools (write your own MCP server in Python or TypeScript)
- Local services running on your machine

If a server exists for it, you can connect it. If one doesn't, you can build one in an afternoon.

## Setup

1. Find or build an MCP server (the [official MCP registry](https://modelcontextprotocol.io) is a good starting point).
2. Run the server (locally, in Docker, or wherever you'd run it).
3. In River, open **Settings → Integrations → MCP Servers → Add Server**.
4. Enter the server's URL or stdio config and any required auth.
5. River fetches the server's tool catalog and registers each tool in your Dataspace.

Once registered, the tools appear in the SpaceChat tool picker and are available to any Agent or Workflow you allow.

## Tool-based by default

MCP integrations are always tool-based — the AI calls the server on demand, the server returns a structured result. There's no continuous sync layer. If you want continuous sync of MCP-sourced data into River entities, build a Workflow that polls the MCP server and writes entities.

## Self-hosted River + local MCP

Pair self-hosted River with local MCP servers and you have a fully on-premises AI workspace. No data leaves your network. No external services involved.

## Tips

- For small one-off tools, prefer a Custom App or a Workflow `http_request` node — MCP is overkill.
- For anything you'll use across many Agents, MCP is the cleanest abstraction.
- The MCP ecosystem moves fast. Check the official registry every few weeks for new servers.

Connect any Model Context Protocol-compatible tool server to River.

MCP Servers

What River is, why it's structured the way it is, and where to start.

Welcome

Get from zero to a working River workspace in five steps.

Quickstart

Plans, AI usage budgets, and how billing works.

Pricing & Plans

The six layers that make River feel like an operating system, not an app.

How River Works

The rooms you work in — persistent, contextual, and returnable.

Spaces

Folders for the modern world — they hold anything, nest infinitely, and know what's inside.

Groups

What's open right now — your current working set on top of everything that's permanent.

Windows

The things themselves — every doc, file, email, task, agent, and artifact in River.

Entities

River's nervous system — every action, human or AI, recorded in real time.

Event Stream

The fastest way back — find any entity by title, content, or context.

Search

Your audit trail — what was created, edited, and produced, in human-readable form.

Activity

How River talks back to you when something matters.

Notifications

The ownership boundary — your personal world model and your team's shared one.

Dataspaces

How AI is layered into River — reactive, scheduled, triggered, and ambient.

Overview

The AI sidebar in every Space — the fastest way to do anything in River.

SpaceChat

The three SpaceChat modes — and what each one guarantees.

Edit, Plan & Ask

How AI acts on entities — the verbs every Agent and Workflow can use.

Tools & Commands

Run AI on your own hardware with Ollama — private, fast, and offline-capable.

Local Models

Reusable AI specialists that run on triggers, schedules, or demand — your autonomous workforce.

Agents

Create, configure, and deploy a River Agent in under five minutes.

Building an Agent

When Agents run — events, schedules, and manual invocation.

Triggers

Focused, single-shot AI capabilities — the specialized moves the AI uses inside larger tasks.

Skills

Deterministic visual graphs for processes that must run the same way every time.

Workflows

Every built-in Workflow node — triggers, data, River-native, and human/LLM-in-the-loop.

Node Types

Recurring automations — like cron jobs, but first-class entities.

Schedules

Every autonomous run, recorded — queued, running, succeeded, or failed.

Jobs

Purpose-built environments for specific kinds of work — all sharing the same entity system and AI.

Apps

Email with AI-powered triage, smart folders, one-click composition, and multi-account support.

Mail

Task management with AI extraction, smart views, and keyboard shortcuts.

Tasks

Fast capture and long-form notes — always in the entity system, always reasonable-over by AI.

Notes

Rich document editing — multiplayer, AI-native, designed for the things people read.

Pages

Reference documents, knowledge bases, and the long memory of your team.

Docs

Spreadsheets with formulas, live data, and AI inside every cell.

Sheets

Your relationship graph — people, companies, and the connections between them.

Contacts

In-workspace web browsing — research, clip, summarize, and save to entities.

Browser

AI image generation and editing with cloud and local providers.

Image Studio

An AI video pipeline — ingest, analyze, edit, and generate.

Video Studio

The control center for Agents, Workflows, and Schedules.

Automations

View, annotate, and extract from PDFs — fully indexed and AI-readable.

Audio and video transcription with speaker detection — searchable, AI-readable, link-ready.

Transcripts

How the world gets in — integrations, the browser extension, the desktop watcher, and direct ingestion.

Senses & Connections

What the River Desktop App unlocks — native files, screen access, and AI that operates your machine.

Desktop & Computer Use

How River connects to the rest of your stack — 100+ services, two patterns.

Integrations

Connect Gmail for full-inbox sync, AI triage, and one-click composition.

Gmail

Sync your calendar as entities the AI can reason over and act on.

Google Calendar

Browse Drive, ingest files as entities, and reason over the contents with AI.

Google Drive

Search Slack, post from River, and receive notifications in your channels.

Clip any web page, selection, or screenshot from anywhere into River.

How organizations, dataspaces, and member access fit together.

Orgs & Members

How access works at the Org, Dataspace, Space, and entity level.

Roles & Permissions

Share individual entities, Groups, or whole Spaces — internally or with guests.

Sharing

Real-time co-editing, cursor presence, and CRDT-based sync that never conflicts.

Multiplayer & Presence

Set your identity, preferences, and AI personalization.

User Profile

Ask the AI for a dashboard, a tracker, or any new view of your data — and have it built on the spot.

Custom Views

Build full applications inside River — your own entity types, UI, data, and AI.

Custom Apps

Pre-configured Spaces for common workflows — one click to spin up a working environment.

Space Templates

Move through River with your hands on the keyboard.

Keyboard Shortcuts

What stays on your machine, what flows to River's cloud, and how you control both.

How Privacy Works

Run the full River stack on your own infrastructure — no cloud dependency.

Self-Hosted River

How team Dataspaces, shared agents, and collaborative workflows work in practice.

Teams Overview

Usage metrics, member management, and org-level controls.

Admin Dashboard

Plans, seats, AI usage budgets, and invoice management.

Billing & Usage

A friendly walkthrough of the six layers and how a single email becomes work.

How River Fits Together

When to use each SpaceChat mode — and what each mode guarantees.

Edit, Plan, and Ask

Why integrations, the browser extension, and computer use are River's perception layer, not just data sources.

The Senses — Perception as a Layer

A practical way to decide between SpaceChat, Skills, Agents, and Workflows.

Automation Mental Model

How information moves from the outside world into River, and where it lands.

Data Flow Basics

How River decides what deserves your attention — before you ask.

Signal Intelligence Basics

How River turns 'I wish I had a tool for this' into 'here, the AI just built it.'

Build Anything — Views, Dashboards, Custom Apps

A compact checklist to orient a new River workspace.

Getting Started

Common questions about River, what it is, and how to use it.

Simple checks when behavior doesn't match expectations.

Troubleshooting Basics

How to think about connected services in River — and unblock the most common issues.

MCP Servers

What MCP unlocks#

Setup#

Tool-based by default#

Self-hosted River + local MCP#

Tips#