Skip to content

OpenClaw

What it is

OpenClaw (formerly Clawdbot/Moltbot) is an open-source, self-hostable autonomous AI agent platform designed for deploying personal and team agents. It runs as a lightweight TypeScript "Gateway" process that interfaces with 50+ messaging channels (Telegram, WhatsApp, Signal, Discord, Slack), executes multi-step workflows, and manages persistent local memory.

What problem it solves

Setting up a personal AI agent that works continuously, remembers context, and integrates with local system resources normally requires complex orchestration. OpenClaw simplifies this by providing a single-port Gateway (18789) that bridges LLMs (GPT-5.2, Claude 4.6, or local models via vLLM) to the user's local operating system and messaging apps.

Where it fits in the stack

Agent Runtime / Orchestration Layer. OpenClaw is the execution environment for autonomous behaviors. It sits between the user's communication channels and the model inference provider (LiteLLM).

┌─────────────────────────────────────────────────────┐
│      User Channels (Signal/WhatsApp/Telegram/CLI)   │
└─────────────────────┬───────────────────────────────┘
                      │
┌─────────────────────▼───────────────────────────────┐
│           OpenClaw Gateway (Port 18789)             │
│  ┌───────────┐  ┌──────────────┐  ┌──────────────┐  │
│  │  Router   │  │  Skill Bus   │  │ Vector Memory│  │
│  └───────────┘  └──────────────┘  └──────────────┘  │
└─────────────────────┬───────────────────────────────┘
                      │
┌─────────────────────▼───────────────────────────────┐
│         LiteLLM / Ollama / OpenRouter                │
│      (OpenAI-compatible model endpoint)              │
└─────────────────────────────────────────────────────┘

Architecture overview

Component Description
Gateway A single TypeScript process managing all connections and tool execution.
Router Receives messages from 50+ channels; selects the appropriate skill or agent loop.
Skill Bus Loads and executes YAML/Markdown skills; enforces sandboxed permissions.
Native Memory (v1.4+) Built-in vector memory layer for semantic recall without external DBs.
Channel adapters Native support for Signal, iMessage, WhatsApp, and 50+ others.
Scheduler Cron-like task runner for timed actions (e.g., "Nightly Research Digest").

Skill system

Skills are self-contained behavior modules defined in YAML. - Triggers: Keyword, regex, slash-command, or cron schedule. - Tools: Native support for shell commands, filesystem access, and browser automation. - Registry: ClawdHub (the community registry) hosts 2,300+ reusable skills.

Typical use cases

  • Personal Assistant: Manage tasks in Vikunja or Home Assistant via chat.
  • Moltbook Engagement: (2026) Autonomous agents participating in AI-only social networks.
  • Local File Automation: Organize downloads, process receipts (OCR), and update local databases.
  • CI/CD Remediation: Automatically analyze build failures and draft PR fixes in GitHub.
  • Scheduled Digests: Aggregate web research and weather into a daily briefing.

Getting started

Installation (macOS/Linux)

OpenClaw is optimized for local execution on macOS (Apple Silicon) and Linux.

# One-command installer (Official 2026 script)
curl -fsSL https://openclaw.io/install.sh | sh

# Start the Gateway
openclaw start

Docker Compose (Self-hosted)

For server environments, use Docker to ensure sandboxed tool execution.

services:
  openclaw:
    image: openclaw/openclaw:latest
    ports:
      - "18789:18789"
    environment:
      GATEWAY_PORT: 18789
      LLM_BASE_URL: "http://litellm:4000"
      LLM_MODEL: "gpt-5.2-mini"
      SIGNAL_SERVICE_URL: "http://signal-api:8080"
    volumes:
      - ./skills:/app/skills
      - ./memory:/app/memory

CLI Reference

OpenClaw provides a powerful CLI for managing the agent:

# Install a skill from ClawdHub
openclaw skill install clawdhub:receipt-processor

# Evaluate agent performance (April 2026 feature)
openclaw eval --suite tests/assistant_bench.yaml

# Inspect the vector memory
openclaw memory query "What did we discuss about the house renovation?"

Hardening and Security (2026 Update)

ClawJacked Vulnerability

Version 2026.2.1 and earlier are vulnerable to a local-gateway authentication flaw. Ensure you are running 2026.2.25 or later.

  • Sandboxing: Always run destructive shell/browser skills in a Docker container.
  • Human-in-the-Loop: Use the confirmation_required: true flag for skills involving financial transactions or external communication.
  • Trusted Sources: Only install skills from verified ClawdHub maintainers; malicious skills can execute arbitrary shell commands.

When to use it

  • For tasks requiring multi-step reasoning and action-taking on a local machine.
  • When you want a ready-to-run personal assistant that works through messaging apps.
  • For home-lab automation tied to Ollama, n8n, Paperless-ngx, or Vikunja.

When not to use it

  • For mission-critical tasks where zero autonomous interpretation is required.
  • If you are uncomfortable maintaining a self-hosted Docker environment.
  • If the target environment cannot support a sandboxed TypeScript process.

Strengths

  • Low Latency: Local Gateway architecture ensures fast tool execution and messaging.
  • Privacy-First: Conversation history and vector memory stay on your local device.
  • Extreme Extensibility: 2,300+ community skills cover almost any API or service.
  • Model Agnostic: Seamlessly switch between Ollama, GPT-5.2, and Claude 4.6 via LiteLLM.

Limitations

  • Security Governance: Requires technical knowledge to properly sandbox and secure.
  • Token Consumption: Autonomous loops can quickly consume API budgets; use LiteLLM to set hard limits.
  • macOS/Linux Focus: Windows support is primarily via WSL2/Docker.

Sources / References

Contribution Metadata

  • Last reviewed: 2026-05-28
  • Confidence: high