cloudparts

n8n

A visual workflow automation tool you can self-host or rent as n8n Cloud: workflows are chains of nodes started by triggers, and its AI Agent node, built on LangChain’s JavaScript library, lets a model call other nodes and MCP servers as tools.

What it is
Workflow orchestration
Used for
Agent frameworks,Agent tools and integrations
Type
Commercial product
Made by
n8n GmbH
License
Sustainable Use License 1.0 (n8n’s own fair-code, source-available license, not OSI open source: use only for your own internal business, personal or non-commercial purposes); files marked .ee. need a paid n8n Enterprise License; Apache 2.0 with Commons Clause until 17 March 2022
Facts checked
13 October 2026
Website Source code

What it is

A workflow starts with a trigger node, such as a webhook, a schedule, a chat message or an event from one of its app integrations, and passes items through action, logic and code nodes that you wire together on a canvas; each run is an execution whose input and output data you can inspect node by node. AI features are cluster nodes built on LangChain’s JavaScript framework: an AI Agent root node takes a chat model sub-node (Anthropic, OpenAI, Ollama and others), optional memory (for example Postgres or Redis chat memory) and at least one tool sub-node, which can be another n8n workflow, a code tool or an MCP server. The MCP Server Trigger node works the other way round, exposing n8n tools and workflows at a URL that MCP clients call over SSE or Streamable HTTP.

A self-hosted instance stores workflows, credentials and executions in SQLite by default or in PostgreSQL. In queue mode, one main instance handles triggers and webhooks and pushes execution IDs to Redis, and any number of worker processes pick them up, load the workflow from the database, run it and write the results back. n8n Cloud prices by full workflow executions, however many steps they have: Starter is 20€ a month billed annually for 2,500 executions and 5 concurrent executions, and Pro is 50€ for 10,000 executions. The code is source-available under the Sustainable Use License, so you may self-host the free Community edition for your own business, but you may not offer n8n as a service, let your customers build or edit workflows through your product, or sell a fork, without a commercial agreement.

Why teams pick it

  • Visual workflows with code when needed: a Code node runs your own JavaScript or Python, and past executions are kept so you can copy their data back into the editor to debug
  • Agents sit inside ordinary automations: an AI Agent node can be preceded by triggers and followed by any of n8n’s app nodes, and other workflows can be its tools
  • Self-hostable for free for internal use, with the source on GitHub, so workflow data and credentials can stay on your own servers
  • Scales out with queue mode (Redis plus workers, each running 10 executions in parallel by default), and Cloud prices per execution rather than per step

Watch out for

  • Not open source: the Sustainable Use License forbids hosting n8n for clients or embedding a workflow builder in your product without a separate commercial license
  • Several team features are paid even when self-hosted: SSO, Git version control, environments, projects, external secrets, log streaming and multi-main mode
  • Queue mode doesn’t work with binary data stored on the filesystem (use S3 external storage instead), and n8n recommends against running it on SQLite
  • Cloud plans cap concurrency: 5 concurrent executions on Starter and up to 50 on Pro

Alternatives

What all workflow orchestration have in common →Compare side by side →
StackStormOpen sourceEvent-driven automation for operations, “IFTTT for ops”: sensors turn alerts and webhooks into triggers, rules match them, and actions or YAML workflows fix the problem, with every run audited.AWS Step FunctionsAWSAWS’s serverless workflow engine: state machines in JSON that call AWS services directly.Cloudflare WorkflowsCommercialDurable execution on Cloudflare Workers: a workflow is ordinary TypeScript or Python code split into steps, and each step’s result is saved so a failed run retries that step and carries on instead of starting over.PrefectOpen sourceA Python workflow orchestrator where a pipeline is an ordinary function decorated with @flow, with retries, caching, schedules and a UI added around it.TemporalOpen sourceAn open-source durable execution engine: you write workflows as ordinary code and they survive crashes.

How it fits with other technologies

Self-hosted queue mode: PostgreSQL for state, Redis for the queue, extra workers · bash

# on the main instance and every worker
export DB_TYPE=postgresdb
export DB_POSTGRESDB_HOST=postgresdb
export EXECUTIONS_MODE=queue
export QUEUE_BULL_REDIS_HOST=localhost
export N8N_ENCRYPTION_KEY=<same key as the main instance>

docker run --name some-redis -p 6379:6379 -d redis

# start a worker that runs up to 5 executions at once (default 10)
n8n worker --concurrency=5

See it in a real stack

No architecture uses n8n yet. Drop it into the playground and wire it up yourself.Open playground →
Where it came from2019–2025

n8n’s GitHub repository dates from June 2019. The company chose what it calls fair-code: the source is public, anyone can use and extend it, but its authors restrict commercial use, because, in n8n’s view, other businesses, especially cloud providers, captured the value of open-source projects with little return to their developers. Until 2022 its license was Apache 2.0 with the Commons Clause, which forbade “selling” the software and also stopped people charging for consulting and support around it.

On 17 March 2022 Jan Oberhauser announced the Sustainable Use License, written in plain English and based, with Elastic’s permission, on the Elastic License 2.0. It narrowed allowed use to internal business purposes and lifted the ban on paid consulting. n8n 1.0 followed in July 2023. n8n 2.0, released as a beta on 8 December 2025, was about secure defaults rather than new features: Code nodes run in isolated task runners by default, and saving a workflow no longer changes the live version until you publish it.

Why it’s built this way

A source-available license of its own instead of open source
n8n wanted the product free to use and widely distributed while keeping the ability to build a business. No existing license fit, so it wrote a short, plain-English one and promoted “fair-code” as an umbrella term for such models.
Replace the Commons Clause
Adding a commercial restriction on top of Apache 2.0 created confusion, and it blocked paid consulting and support, which n8n saw as critical to helping people get value from the product.
Secure defaults in 2.0, at the cost of breaking changes
Under semantic versioning, the hardening n8n wanted (isolated Code node execution, blocked environment variables, disabled command-execution nodes) had to wait for a major version, so 2.0 shipped them together.

Timeline

  1. 2019

    n8n repository created on GitHub

  2. 2022

    License changes from Apache 2.0 with Commons Clause to the Sustainable Use License

  3. 2023

    n8n 1.0 released

  4. 2025

    n8n 2.0 beta: task runners on by default, separate Save and Publish

Sources13 · facts checked 13 October 2026

Improve this page on GitHub