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

# Introduction

> A security-first personal AI agent that runs on serverless infrastructure.

# Lite Claw

A personal AI assistant that prioritizes security over convenience. Unlike traditional AI agents that expose network ports and run with full system access, Lite Claw uses a locked-down architecture with no attack surface.

<CardGroup cols={2}>
  <Card title="No Exposed Ports" icon="shield-halved">
    Outbound-only connections to Telegram. Nothing to find on Shodan.
  </Card>

  <Card title="Whitelist Authentication" icon="user-check">
    Only approved Telegram user IDs can interact with your agent.
  </Card>

  <Card title="Confirmation Gates" icon="lock">
    Write operations require explicit approval with random codes.
  </Card>

  <Card title="Serverless Native" icon="cloud">
    Scales to zero. Pay only when running. \~\$5-15/month.
  </Card>
</CardGroup>

## What can it do?

Lite Claw connects to your productivity tools and lets you interact via natural language in Telegram:

* **Calendar** — Read and create Google Calendar events
* **Email** — Summarize Gmail inbox
* **Weather** — Daily forecasts
* **Scheduled briefings** — Morning summaries and weekly reviews

```
You: What's on my calendar tomorrow?

Bot: Tomorrow (Monday, Feb 10):
     • 9:00am - Standup [Work]
     • 11:30am - Lunch with Alex [Personal]
     • 3:00pm - Design review [Work]

You: Schedule dentist Friday 2pm

Bot: I'll create this event:
     Title: Dentist
     When: Friday, Feb 14, 2:00pm

     Reply YES 482193 to confirm.

You: YES 482193

Bot: Done! Added to your calendar.
```

## Who is this for?

Lite Claw is for developers who want a personal AI assistant but don't trust the security posture of existing solutions. If you've looked at OpenClaw and thought "I'm not exposing port 18789 to the internet," this is for you.

**You should use Lite Claw if:**

* You want an AI assistant you control
* Security matters more than features
* You're comfortable deploying to Railway/Vercel/Lambda
* You want to pay for usage, not subscriptions

## Get started

<CardGroup cols={2}>
  <Card title="Quickstart" icon="rocket" href="/quickstart">
    Deploy to Railway in 10 minutes.
  </Card>

  <Card title="Security Model" icon="shield" href="/security">
    Understand the threat model and defenses.
  </Card>
</CardGroup>

## Stack

| Component | Technology                           |
| --------- | ------------------------------------ |
| Runtime   | TypeScript on Railway (long polling) |
| Database  | Neon Postgres                        |
| Cache     | Upstash Redis                        |
| LLM       | Claude via OpenRouter                |
| Interface | Telegram Bot API                     |