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

# OpenClaw Integration

> Deploy OpenClaw as a hosted cloud agent on the Neosantara dashboard or run locally via the official provider plugin.

[OpenClaw](https://openclaw.ai) is an autonomous multi-channel personal AI agent designed to execute commands, automate tasks, and coordinate workflows across [Telegram](https://telegram.org/?utm_source=neosantara-docs\&utm_medium=referral), Discord, Slack, and terminal environments. You can run OpenClaw using two primary modes: **Hosted Cloud Agent** (zero-maintenance deployment in the Neosantara dashboard) or **Self-Hosted CLI** using the `@neosantara/openclaw-provider` plugin.

## Deployment Options

| Option                 | Execution Environment   | Setup Requirements      | Primary Use Case                                                    |
| :--------------------- | :---------------------- | :---------------------- | :------------------------------------------------------------------ |
| **Hosted Cloud Agent** | Cloud E2B microVM       | Telegram bot token only | 24/7 personal assistant on Telegram with automated scale-to-zero.   |
| **Self-Hosted CLI**    | Local machine or server | Node.js & OpenClaw CLI  | Local terminal coding, file system modifications, and custom tools. |

***

## Option 1: 1-Click Hosted Cloud Agent (Dashboard)

The fastest way to deploy OpenClaw without maintaining server infrastructure is through Neosantara's managed cloud runtime.

### 1. Get a Bot Token and Owner ID

1. Open [@BotFather](https://t.me/BotFather) on Telegram, send `/newbot`, and copy the generated **Bot Token**.
2. Open [@userinfobot](https://t.me/userinfobot) on Telegram to find your numerical **User ID**.

### 2. Deploy from the Dashboard

1. Navigate to [app.neosantara.xyz/cloud/openclaw](https://app.neosantara.xyz/cloud/openclaw).
2. Enter your **Bot Token** and **Telegram Owner ID**.
3. Select a primary model (such as `deepseek-v4.1-flash`) and an optional fallback model.
4. Click **Deploy Agent**.

The isolated E2B microVM boots in \~700ms whenever an incoming Telegram message arrives, and automatically pauses after 30 minutes of inactivity with zero compute cost.

***

## Option 2: Self-Hosted CLI & Terminal

To run OpenClaw directly on your local workstation or VPS, install the official Neosantara plugin.

### 1. Install Neosantara Provider Plugin

Run the following command in your terminal:

```bash theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
openclaw plugins install clawhub:@neosantara/openclaw-provider
```

### 2. Configure Credentials

Set your Neosantara API key via environment variables:

```bash theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
export NEOSANTARA_API_KEY="nsk_your_api_key_here"
```

Alternatively, supply the key during interactive onboarding:

```bash theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
openclaw onboard --neosantara-api-key "nsk_your_api_key_here"
```

### 3. Select a Model

Set OpenClaw to route through the gateway using the `neosantara/` namespace:

```bash theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
openclaw model set neosantara/deepseek-v4.1-flash
```

### 4. Run OpenClaw

Launch the daemon gateway or enter interactive terminal mode:

```bash theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
# Start the background daemon
openclaw gateway

# Launch the interactive terminal interface
openclaw tui
```

***

## Recommended Models

| Use Case                          | Model Identifier      | Context Window | Characteristics                                                         |
| :-------------------------------- | :-------------------- | :------------- | :---------------------------------------------------------------------- |
| **Reasoning & Autonomous Coding** | `deepseek-v4.1-flash` | 1M tokens      | High tool calling accuracy for complex shell commands and file changes. |
| **Low Latency & Fast Turnaround** | `gemini-3.8-flash`    | 1M tokens      | Minimal response latency suited for interactive Telegram messaging.     |
| **Cost-Sensitive Tasks**          | `gpt-5.4-mini`        | 128k tokens    | Budget-friendly option for routine summaries and daily alerts.          |

## Example Command Execution

Once connected via Telegram or TUI, assign autonomous tasks to OpenClaw:

```text theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
> Check open issues in our GitHub repository, summarize top priorities, and draft an update.
```

OpenClaw queries the Neosantara gateway, executes relevant command tools or web lookups, and returns structured findings.

## Next Steps

| Goal                            | Guide                                         |
| :------------------------------ | :-------------------------------------------- |
| Hosted cloud agent architecture | [Cloud Agents Guide](/en/agents/cloud-agents) |
| Hermes Agent deployment         | [Hermes Agent Guide](/en/agents/hermes)       |
| Terminal coding with Pi         | [Pi Coding Agent Guide](/en/agents/pi)        |
| MCP tool connector setup        | [MCP Connector](/en/agents/mcp-connector)     |


## Related topics

- [Hosted Cloud Agents](/en/agents/cloud-agents.md)
- [Hermes Agent Integration](/en/agents/hermes.md)
- [MCP & Agent Architecture](/en/agents/overview.md)