Post-Installation Setup
After installing OpenClaw, complete onboarding: choose your LLM provider, add your first messaging channel (Telegram is easiest), test the connection, and run through the security hardening checklist. This guide walks you through each step.
1. Overview
Post-installation setup ensures your OpenClaw instance is usable and secure. You will:
- Run the onboarding wizard (or equivalent first-run flow) to select an LLM and add a channel
- Configure your first channel - Telegram is recommended for first-time users
- Test the connection by sending a message to your agent
- Optionally set up dashboard access and review security hardening
Time needed: about 5–15 minutes depending on whether you already have API keys and a messaging app account.
2. Prerequisites
- OpenClaw installed and runnable - You’ve completed one of: Quick Start, macOS, Windows, Linux, or Docker. Verify with
openclaw status. - LLM API key - At least one provider: Anthropic (Claude), OpenAI (GPT), or optionally Google (Gemini), DeepSeek, Ollama (local).
- Messaging app account - For your first channel: Telegram (recommended), Discord, or WhatsApp. See Channel Setup Hub.
- Basic terminal/CLI - You should be comfortable running commands in a terminal.
3. Step-by-Step Post-Installation
Step 1: Verify installation
Confirm OpenClaw is on your PATH and the gateway can run:
openclaw status
If you see “command not found”, fix your PATH or re-run the installer. For deeper checks, run:
openclaw doctor
See Quick Diagnostics and Troubleshooting Guide if anything fails.
Step 2: Run the onboarding wizard
The first time you run OpenClaw, an onboarding wizard (or interactive setup) typically guides you through:
- Selecting an LLM provider - Choose Anthropic Claude, OpenAI GPT, or another supported provider. You’ll be prompted to enter your API key (or path to it). Store keys in environment variables when possible; see Credential Management.
- Adding your first channel - You’ll configure one messaging platform (e.g. Telegram bot token, Discord bot, WhatsApp). Telegram is the fastest to set up for most users.
If your install uses a config file instead of a wizard, edit the config: openclaw config edit (or edit the config file directly). Add a model provider and a channel section. Structure details: Configuration file guide.
Step 3: Configure your first channel (Telegram recommended)
For the first channel, Telegram is recommended: create a bot via @BotFather, get the token, and add it in the wizard or config. Full steps: Telegram setup guide. Alternatives: Discord, WhatsApp, Slack, and more in the Channel Setup Hub.
Step 4: Test the connection
Start the gateway if it’s not already running, then send a message to your bot (e.g. in Telegram). The agent should reply. If it doesn’t:
- Check
openclaw statusand that the gateway is running - Confirm the channel is connected (see CLI or config)
- Check logs:
openclaw logs --follow(or equivalent) for errors - Verify API key and model provider; see Channel errors and Model & API errors
Step 5: Dashboard access (optional)
If your deployment supports a web dashboard, configure access according to the official docs or your install method. Keep the dashboard behind authentication and only on localhost or a secure tunnel (e.g. SSH, Tailscale/VPN). Never expose the gateway or dashboard publicly without proper security.
Step 6: Security hardening checklist
Before treating the instance as “done”, run through our Security Checklist and Security Best Practices. Essential items:
- Gateway bound to
127.0.0.1only (not 0.0.0.0) - API keys in environment variables or a secrets manager; file permissions 600 for any key files
- API spending limits set where supported
- Firewall and network isolation if the machine is reachable from the internet
- Sandbox or tool restrictions enabled if you’re concerned about agent capabilities
Details: Security Checklist, Best Practices, Credential Management, Network Isolation.
4. Common Post-Installation Issues
| Issue | Solution |
|---|---|
| Connection failed / channel not receiving | Check bot token and channel config; ensure gateway is running; see Channel errors and Gateway issues. |
| Authentication or API key error | Verify API key is valid and has correct scope; use env vars; see Credential Management and Model & API errors. |
| Gateway won’t start | Port 1618 (or configured port) free; bind to 127.0.0.1; run openclaw doctor; Gateway troubleshooting. |
| Config validation failed | Check schema and required fields; Configuration file guide and Configuration problems. |
| Agent not responding to messages | Confirm channel is connected, model provider is set, and logs for rate limits or model errors; Channels, Models. |
| Performance or slow responses | Check system resources, network latency to LLM API, and log volume; Performance problems. |
Run openclaw doctor first when in doubt. Full list: Troubleshooting Guide.
5. Best Practices
- Security first - Follow security best practices and the security checklist; bind gateway to localhost; use spending limits.
- Updates - Keep OpenClaw updated for security and features; test in a dev environment when possible.
- Logs and monitoring - Periodically check logs for errors or unusual activity; see Monitoring & Logging.
- Backups - Back up configuration and any custom skills or data.
- Skills - Install skills from ClawHub only after reviewing; run security audit and pin versions.
6. Next Steps
- Add more channels (Discord, WhatsApp, Slack, etc.).
- Customize agent settings and configuration.
- Install skills from ClawHub and browse top 50 skills.
- Explore use cases (productivity, business, content, development).
- Complete the security checklist and join the OpenClaw community (GitHub, Discord).
7. Related Resources
- Official Documentation - Install and config reference
- Video Tutorials - First-run and channel setup
- Configuration Templates - Example configs
- Configuration File Guide - Config layout and options