n8n-claw
Open Source & Self-Hosted

Your AI agent.
Your infrastructure.

A fully self-hosted AI agent (OpenClaw alternative) built on n8n. Talks to you via Telegram, Slack, Teams or own app, builds its own tools via skills, manages tasks and memory — all running on your own server. No cloud lock-in.

19+
Built-in Tools
100%
Self-Hosted
10 min
Setup time
What n8n-claw is and how it works
Everything you need in one agent
Talk to your agent in natural language. It manages tasks, remembers context, builds integrations, and keeps you on track — autonomously.
💬

Telegram & Webhook API

Chat via Telegram or connect any system through the HTTP API — Slack, Teams, or your own apps. One agent, many interfaces.

🧠

Memory, Knowledge Graph & User Modeling

Hybrid search (semantic + full-text + entity, fused via RRF), time decay, and a knowledge graph of people and companies. Nightly pattern extraction turns conversations into a behavior model — your agent learns how you work, not just what you say.

🎭

Custom Personality

Give your agent a name, choose its language, set the tone — casual, professional, or fully custom. It's your assistant, shaped to fit how you work.

Task Management

Create, track, and complete tasks with priorities and due dates. The agent proactively reminds you of overdue items.

Smart Reminders

"Remind me in 2 hours to call back." One-time reminders, scheduled actions, and timed instructions — create, list, edit, or delete them in natural language.

🔄

Recurring Actions

"Check my emails every 15 minutes", "Daily briefing at 8am" — schedule any repeating task in natural language.

🤖

Expert Agents

Delegate complex tasks to specialized sub-agents. 3 included out of the box, expandable from the agent catalog.

🔧

MCP Skills

Install pre-built API integrations from the skill catalog — or tell the agent to build new ones from scratch. In natural language.

📧

Google Services

Connect Gmail, Google Calendar, Analytics, and Ads via OAuth2. The agent handles the entire auth flow via Telegram — no manual config needed.

🌐

Web Search & Reader

Built-in web search via SearXNG (no API key needed) and clean webpage reading via Crawl4AI. Your agent browses for you.

🎤

Voice, Photos, Docs & Files

Send voice messages, photos, PDFs, or your location. Voice is transcribed, photos are analyzed, documents are extracted. The agent can also download files from the internet or cloud services and send them back to the chat.

🔍

Smart Background Checks

Monitoring tasks that run silently and only notify you when something new is found. No noise, just signal.

📁

Project Memory

Persistent markdown documents for tracking ongoing work across conversations. Your agent never loses the thread.

🖥️

OpenClaw Integration

Delegate coding tasks to an autonomous AI agent. Your assistant sends the task, OpenClaw writes the code, builds the app, and reports back.

🔀

Any LLM Provider

8 providers out of the box: Claude, GPT, Gemini, Mistral, DeepSeek, Ollama, OpenRouter, or any OpenAI-compatible endpoint. Choose during setup — no manual config needed.

65+ ready-to-install integrations
Say "install the HubSpot skill" in chat — and your agent can create contacts, update deals, and log activities. No coding, no config files.
🎯

CRM

HubSpot, Salesforce & Zoho — contacts, deals, pipelines

📋

Project Management

Jira, Confluence & Asana — issues, sprints, pages, tasks

💳

Payments & Invoicing

Stripe & Lexware — subscriptions, invoices, bookkeeping

🏦

KontoFlux Banking

Access 5000+ German bank accounts (read-only)

🗄️

Smart Spreadsheets

Airtable & NocoDB — records, views, automations, self-hosted option

📝

Notion & Todoist

Search pages, manage tasks, sync projects

📧

Gmail & Email

Read, send, search emails via Gmail API or IMAP/SMTP

📅

Calendars

Google Calendar & CalDAV — events across any calendar provider

🎙️

Vexa Meeting Bot

Auto-join Google Meet, Teams & Zoom — live transcription

📊

Marketing Analytics

Google Ads & GA4 — campaigns, spend, conversions (Meta coming)

🏠

Home Assistant

Read states, trigger services — lights, scenes, automations

And 50+ more…

GitHub, Nextcloud, DeepL, YouTube, Wikipedia, Hacker News, Vikunja, recipes…

100+ specialized sub-agents
Your agent knows when to call in a specialist. Say "analyze my sales pipeline" and it delegates to a dedicated Pipeline Analyst agent.
📊

Analytics 9 agents

Data analysis, conversion tracking, ads performance, financial modeling, KPI dashboards

📣

Marketing 16 agents

SEO audits, social media strategy, content creation, email campaigns, growth hacking

💻

Development 25 agents

Backend, frontend, DevOps, security audits, QA testing, code reviews, architecture

💹

Sales 8 agents

Outbound strategies, pipeline analysis, proposal writing, partnership development

🎨

Creative 8 agents

UX design, video production, podcast planning, brand identity, visual storytelling

⚙️

Operations 14 agents

Customer support, project management, legal review, event planning, process optimization

🎯

Product 6 agents

Product strategy, feature prioritization, user research, competitive analysis

🔬

Research 5 agents

Market research, academic analysis, competitive intelligence, UX research

Up and running in minutes
One command to install. Interactive setup guides you through everything. No Kubernetes, no YAML nightmares.

Clone & Run

One command on your VPS or local machine. The setup script handles Docker, databases, credentials, and all services.

git clone https://github.com/freddy-schuetz/n8n-claw.git && cd n8n-claw && ./setup.sh

Configure

Interactive prompts ask for your Telegram bot token, LLM provider, and agent personality. Choose casual, professional, or fully custom.

Chat

Send a message to your Telegram bot. Your personal AI agent is live — all credentials and workflows are set up automatically.

What your agent does for you
You just chat. Your agent handles everything else — automatically, in the background, 24/7.
💬

You talk — it acts

Send a message via Telegram. Your agent understands what you need, takes action, and reports back. No buttons, no menus — just conversation.

🧠

It remembers everything

Had a conversation last week? Your agent still knows the context. It builds up knowledge over time and gets more helpful the longer you use it.

It works while you sleep

Background checks, recurring tasks, daily summaries — your agent keeps running even when you're offline. It only pings you when something needs your attention.

🔌

It connects your tools

Need to check a website, call an API, or search the web? Your agent has built-in skills — and can learn new ones on demand, just by asking.

🤝

It delegates to experts

Complex task? Your agent spins up specialized sub-agents to handle it — like a team lead who knows when to bring in a specialist.

🏠

It stays on your server

No cloud service. No subscription. Everything runs on your own machine. You own every conversation, every memory, every piece of data.

🔒 Your Data. Your Server. Period.

n8n-claw runs entirely on your infrastructure. No data leaves your network except LLM API calls. Perfect for GDPR compliance and privacy-conscious users.

🏠 100% Self-Hosted
🇪🇺 GDPR-Ready
🔑 You Own Your Data
📜 Open Source (MIT)
Built on proven tools
No exotic dependencies. Industry-standard components you already know and trust.
n8n
🐘 PostgreSQL
🧠 Any LLM (Claude, GPT, …)
🗃️ Supabase
🐳 Docker
🔍 SearXNG
🕷️ Crawl4AI
🔗 PostgREST
🌐 Kong API Gateway
📡 Telegram Bot API
I wanted an AI assistant that truly belongs to me — running on my server, with my data, under my control. n8n-claw makes that real.
— Friedemann Schütz, Creator & n8n Ambassador
Friedemann Schütz
Friedemann Schütz

Automation Expert & n8n Ambassador from Essen, Germany. I help teams automate workflows with n8n, integrate AI into their processes, and run self-hosted infrastructure — no cloud lock-in, full data ownership.

🌐 friedemann-schuetz.de 💼 LinkedIn 📅 Book a call

Ready to run your own AI agent?

One command. Five minutes. Full control.