> ## Documentation Index
> Fetch the complete documentation index at: https://clawhqplatform.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# FAQ

> Common questions about ClawHQ.

## General

**Is ClawHQ really free?**
Yes. ClawHQ is MIT licensed and self-hosted — you pay nothing to us. You do pay your AI provider (Anthropic, OpenAI, etc.) for API usage. The Model Router helps minimize those costs by routing cheap tasks to cheaper models automatically.

**What's the difference between ClawHQ and just using ChatGPT?**
ChatGPT is a chat interface. ClawHQ is an agent platform — agents run autonomously, respond to events, execute tools, and work in the channels your team already uses (Discord, Slack, Telegram). They don't need you to start a conversation.

**Do I need technical skills to run this?**
You need to be comfortable with Docker and a terminal. If you can run `docker compose up`, you can run ClawHQ. The `install.sh` script handles most of the setup.

**Can I use models other than Claude?**
Yes. ClawHQ supports Anthropic, OpenAI, Groq, OpenRouter, DeepSeek, and local models via Ollama. Set your API keys in **Settings → General** and configure fallbacks in the Model Router.

## Packs

**How much do packs cost?**
Free packs (Data Analyst, Weekly Briefing, Content Engine, Innovation Scout, Workflow Docs, LLM Wiki) are $0 — install and use forever. Paid packs have three tiers: Decision Brief at $29, lighter workflow packs (Creator Suite, Reskilling Core, Marketing Agency, Product Research) at $49, and heavy multi-agent packs at $99. Bundles start at \$79.

**What's the difference between the $49 and $99 packs?**
The $49 packs are focused single-workflow tools designed for individuals or small teams. The $99 packs involve heavier multi-agent orchestration — multiple specialized agents coordinating across tools like CRMs, email, calendars, and external APIs.

**Are there bundles?**
Yes. Four thematic bundles are available at a discount: Sales Bundle ($149), Finance Bundle ($149), Engineering Bundle ($149), and Starter Bundle ($79). An All Packs Bundle (\$299) includes every paid pack. See [clawhqplatform.com/packs](https://clawhqplatform.com/packs) for details.

**Can I buy a pack and install it on multiple servers?**
Each pack license is per-workspace. If you're running multiple ClawHQ instances (e.g., staging + production), reach out and we'll sort it out.

**What's the refund policy?**
Packs are digital products. If something doesn't work as described, contact us and we'll make it right.

## Hosting

**Where should I host this?**
Any VPS with 2GB+ RAM works. Popular options: Railway (one-click deploy), Render (free tier available), DigitalOcean (\$6/mo Droplet), Hetzner (cheap EU), your own server.

**Can I run it locally?**
Yes — `docker compose up` works on your laptop. Use `docker-compose.local.yml` for local dev (skips HTTPS). Note: Discord and Telegram webhooks require a publicly accessible URL; use ngrok or Cloudflare Tunnel for local testing with live channels.

**Does it support HTTPS?**
Yes. The included Caddy reverse proxy handles TLS automatically. Set `CLAWHQ_URL` to your domain and Caddy will get a certificate from Let's Encrypt.

## Agents

**How many agents can I run?**
As many as your server can handle. Each agent is lightweight — a 2GB VPS can comfortably run 10+ agents. The main cost is API calls, not compute.

**Can agents talk to each other?**
Yes. Via Paperclip orchestration, agents can delegate tasks to other agents. See the [gstack task pipeline](/guides/packs) for structured multi-agent workflows.

**Do agents remember past conversations?**
Each agent has persistent memory (stored in SQLite). Memory search is on by default — agents automatically retrieve relevant past context before responding. You can configure memory settings per-agent.

## Security

**Is my data sent to Anthropic?**
Your messages are sent to whichever AI provider you configure (Anthropic, OpenAI, etc.) for model inference. ClawHQ itself never stores or transmits your data to any third party — it's self-hosted.

**Can I use it without any cloud AI providers?**
Yes — with Ollama. Enable **Local Ollama Routing** in the Model Router and install Ollama on your server. Agents will run fully locally with no external API calls.

**How is the dashboard secured?**
WorkOS authentication is built in — supports email magic links, social login (Google, GitHub), and enterprise SSO (SAML/SCIM). The dashboard is not accessible without signing in.