n8n-claw

n8n-claw is a fully self-hosted AI agent built on n8n. It is presented as an OpenClaw alternative and is meant to run on a user’s own server, with no cloud lock-in. The agent talks through Telegram, Slack, Teams, or a custom app, and it handles tasks and memory while working autonomously.

Its feature set includes natural-language interaction, task management, reminders, recurring actions, and persistent memory. The memory system is described as using hybrid search across semantic, full-text, and entity search, along with time decay and a knowledge graph of people and companies. It also uses nightly pattern extraction to turn conversations into a behavior model. The agent can be given a name, language, and tone, and it can create, track, and complete tasks with priorities and due dates. It also sends proactive reminders for overdue items.

n8n-claw supports web access and integrations through its chat interfaces and HTTP API. The page says it can connect any system through the webhook API, and it can install pre-built API integrations from a skill catalog or build new ones from scratch in natural language. Built-in examples include Google services such as Gmail, Google Calendar, Analytics, and Ads via OAuth2, plus web search with SearXNG, page reading with Crawl4AI, and browser use with Chromium. It also accepts voice messages, photos, PDFs, locations, and files, and can perform background monitoring tasks, maintain project memory in markdown documents, and delegate work to specialized sub-agents. The page also lists support for multiple LLM providers, including Claude, GPT, Gemini, Mistral, DeepSeek, Ollama, OpenRouter, and any OpenAI-compatible endpoint.

The product is described as open source and self-hosted, with 19+ built-in tools, 70+ ready-to-install integrations, 100+ specialized sub-agents, and a 10 minute setup time. It is aimed at users who want an AI assistant running on their own infrastructure.