n8n-io/n8n
Project Overview
n8n
Fair-code workflow automation platform with visual node editor, custom code, and native AI agent capabilities.
True Intent (AI Assessment)
A production-grade, self-hostable workflow automation and AI agent orchestration platform with a visual editor and 400+ integrations.
README Accuracy
Notes
README claims 1500+ integrations while topics say 400+; likely counts community nodes vs. built-in. License is source-available (fair-code), not OSS—users should note the Sustainable Use License restricts hosting-as-a-service.
n8n is a large TypeScript monorepo implementing a self-hostable workflow automation platform. It combines a Vue-based visual editor (drag-and-drop node canvas) with a Node.js backend that executes workflows composed of pre-built integration nodes or custom JavaScript/Python code. The repository is organized as a pnpm workspace with turbo orchestration and includes hundreds of packages under packages/, most notably cli (the main server), core (execution engine), workflow (shared types), editor-ui (frontend), and nodes-base (400+ integration nodes).
The problem it solves is connecting disparate SaaS APIs, databases, and AI models without writing bespoke integration glue code. Users describe multi-step processes visually—triggers, HTTP calls, transformations, conditionals, and now AI agent steps backed by LangChain—rather than maintaining custom scripts. Recent versions heavily emphasize AI: the @n8n/n8n-nodes-langchain package integrates LLM providers (OpenAI, Anthropic, Google), vector stores, and MCP (Model Context Protocol) client/server support.
Users range from individual developers running `npx n8n` locally, to teams self-hosting via Docker, to enterprises on n8n Cloud. Typical usage: define a trigger (webhook, cron, event), chain integration nodes, optionally embed JS/Python code or an AI agent, then deploy. Enterprise features (SSO, RBAC, audit logs) live under a separate EE license.
Architecturally, the CLI package boots an Express server backed by TypeORM (SQLite/Postgres), queues jobs (BullMQ-style workers), and exposes a REST API consumed by the Vue editor. The core package handles node execution, credentials, and binary data (with S3/Azure Blob support). A task-runner subprocess isolates user code execution. The build system uses turbo, biome, and vitest; heavy tooling includes SBOM generation, license compliance checks, and extensive Claude/agent skill definitions in .agents and .claude directories.
Notable observations: the codebase is unusually mature and well-instrumented (Sentry, code-health baselines, mutation testing, telemetry). Licensing is 'fair-code' via the Sustainable Use License plus a separate Enterprise License—source-available but not OSI open source. The .claude/plugins and .agents/skills directories indicate deep internal use of AI coding assistants for development workflow.
| Languages | TypeScript, Vue, JavaScript, SCSS, Python |
| Runtime | Node.js >=22.22 |
| Framework | Express (backend), Vue 3 (frontend), LangChain (AI) |
| Database | SQLite (default), PostgreSQL (via TypeORM) |
| Package Manager | pnpm >=10.22 (workspace monorepo) |
| Key Dependencies | @langchain/core, TypeORM, Express, Vue, @aws-sdk/client-s3, @sentry/node, zod, oclif |
| Build Tool | Turbo + TypeScript compiler (tsc) + Vite |
| Test Framework | Vitest (unit/integration), Playwright (E2E) |