n8n Workflow Automation Platform

repository·master·Indexed 11 days ago

https://github.com/n8n-io/n8n

A fair-code workflow automation platform for building and operationalizing AI agents and complex workflows. It features a visual node-based interface, support for JavaScript and Python custom code, and connectivity to over 1500 services. Includes documentation on Docker deployment, PostgreSQL integration, isolated task runners for code execution, and developer utilities for creating vector store nodes.

Tokens
414.1K
Snippets
1.1K
Records
1.8K
Agent score
97%

What's inside n8n

  1. Overview of @n8n/blob-storage

    master

    The @n8n/blob-storage package provides the blob storage layer for n8n. It is used to persist large blobs, such as execution data and binary data, outside of the primary database.

    To keep the core package lightweight, the root entrypoint does not include cloud SDKs. Instead, it provides types, configurations, and stream utilities. Specific storage implementations are provided via subpath exports and are designed to be loaded dynamically using await import() to ensure cloud SDKs (like S3 or Azure) are only loaded when external storage is actually configured.

  2. Overview of n8n capabilities

    master

    n8n is a fair-code platform designed for building and deploying AI agents and workflow automation.

    Key features include:

    • AI-Native Automation: Build multi-step agents using your own data, models (OpenAI, Anthropic, Google, or open-source), and tools.
    • Hybrid Building: Combine a visual canvas with custom JavaScript, Python, and npm packages.
    • Extensibility: Connect to over 1500 integrations and use 9,000+ workflow templates.
    • Deployment Options: Run it self-hosted or via n8n Cloud.
    • Enterprise Features: Supports role-based access, audit trails, and secure handling of sensitive data.
  3. Overview of @n8n/frontend-utils

    master

    @n8n/frontend-utils is a collection of framework-light utility functions used across n8n's front-end packages. These utilities are designed to be stateless and pure, meaning they are not coupled to Vue components or specific application stores.

    Key characteristics include:

    • Pure utilities: Stateless helpers with no component or store coupling.
    • Consistency: Provides a single source of truth for cross-package front-end utilities.
    • Extensible: Designed to serve as a foundation for growing front-end utility needs.
  4. Overview of Computer-use evaluation

    master

    The Computer-use evaluation is a suite of auto-runnable scenarios designed for testing the Instance AI computer-use feature. It is intended for the inner loop of system-prompt tuning, providing fast feedback against a real local n8n instance without requiring a LangSmith dependency.

    The evaluation targets four specific failure modes:

    1. Failure to propose computer-use: Detected via trace.mustCallMcpServer.
    2. Looping or tool-call budget exhaustion: Detected via trace.mustNotLoop and trace.budget.
    3. Context ballooning: When a single tool result (like a browser_snapshot) consumes excessive tokens, detected via trace.budget with token caps.
    4. End-to-end task failure: Detected via filesystem graders like fs.fileMatches and fs.fileExists.
  5. Overview of Browser MCP

    master
    Browser MCP is a Model Context Protocol (MCP) server that enables AI agents to control a Chrome browser. It connects to the user's local Chrome installation via the n8n Browser Bridge extension. This allows the agent to use the user's actual profile, cookies, and login sessions to perform actions like navigating pages, clicking elements, filling forms, and executing JavaScript.
  6. What is @n8n/local-gateway?

    master
    The @n8n/local-gateway is a native tray application that acts as a bridge between an n8n instance (cloud or self-hosted) and your local machine. It allows n8n workflows to perform local actions such as taking screenshots, controlling the mouse/keyboard, executing shell commands, or interacting with the local filesystem. It runs silently in the system tray and connects to n8n using a gateway token via an OS deeplink.
  7. Overview of @n8n/stores

    master

    @n8n/stores is a collection of Pinia stores designed to provide common data-related functionality across n8n's various Front-End packages. It centralizes state management to ensure consistency and reusability.

    Key Features

    • Composable State Management: Allows sharing and reusing stateful logic across multiple Vue components via Pinia stores.
    • Consistent Patterns: Enforces uniform state handling and best practices across the entire front-end codebase.
    • Composition API Support: Built to work seamlessly with Vue's Composition API for modern development workflows.
    • Extensibility: Designed to support scalable development by allowing new stores to be added or existing ones modified as requirements evolve.
  8. Overview of the n8n CLI

    master

    The n8n CLI is a lightweight terminal client designed to manage n8n workflows, executions, credentials, and other platform resources. It is built to be highly scriptable and compatible with modern development workflows.

    Key Features:

    • Zero Dependencies: Uses native Node.js fetch, ensuring fast installation without heavy HTTP libraries or server stacks.
    • AI Agent Friendly: Designed for seamless interaction with AI coding agents (like Claude Code or Cursor) by providing self-documenting --help flags at every command level.
    • Unix Toolchain Integration: Supports multiple output formats including JSON, table, and ID-only, allowing you to pipe data into tools like jq, grep, or xargs.
    • CI/CD Ready: Follows standard Unix conventions with meaningful exit codes, error messages sent to stderr, and data sent to stdout.
  9. Overview of the Log Streaming Module

    master

    The Log Streaming module is an enterprise-grade feature (feat:logStreaming license required) that forwards n8n internal events to external destinations in real-time. It enables centralized logging, monitoring integration, and audit trails.

    Key Capabilities:

    • Real-time streaming: Forward events to external systems immediately.
    • Multiple destinations: Supports Webhook, Sentry, and Syslog out-of-the-box.
    • Resilience: Uses a circuit breaker pattern to prevent cascading failures.
    • Filtering: Subscribe to specific events using glob patterns.
    • Anonymization: Optional payload anonymization for sensitive data.
    • Scalability: Multi-instance coordination via pub/sub across main, worker, and webhook processes.
  10. Overview of the Workflow Evaluation Framework

    master

    The Workflow Evaluation Framework tests whether workflows built by Instance AI function correctly by executing them using LLM-generated mock HTTP responses. This approach ensures no real credentials or external services are required during testing.

    The framework provides five distinct harnesses for different testing needs:

    • eval:instance-ai: End-to-end testing that includes building, mocked execution, and LLM verification by driving a running n8n instance.
    • eval:agents: Evaluates intent-resolution cases (grading plain build requests based on enacted routing behavior) from the LangTracer agents suite.
    • eval:subagent: A legacy harness for workflow-build compatibility, driving the live orchestrator/skill build path with binary scoring.
    • eval:discovery: An in-process orchestrator test that scores against required or forbidden tool/dispatch events without requiring an n8n server.
    • eval:pairwise: Compares live orchestrator workflow builds using an LLM judge panel against specific do/don't lists, intended for head-to-head comparison with ai-workflow-builder.ee.
    • eval:computer-use: Grades the computer-use agent (handling file, OAuth, or doc-reading tasks) against specific fixtures.