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

# OpenClaw

> Add long-term memory to OpenClaw with the Nebula plugin

Long-term memory for [OpenClaw](https://docs.openclaw.ai) powered by Nebula. Automatically captures conversations and builds a [vector graph](/guides/concepts#the-vector-graph) that the agent can search to recall semantics, procedures, episodes, and sources from past interactions.

<Info>You'll need a Nebula API key and collection ID before starting. Get both at [zeroset.com](https://zeroset.com)</Info>

## Installation

```bash theme={null}
openclaw plugins install @nebula-ai/openclaw-nebula
```

Restart OpenClaw after installation.

## Configuration

<Tabs>
  <Tab title="Environment Variables">
    The fastest way to get started. Add your credentials to the OpenClaw environment:

    ```bash theme={null}
    echo -e "NEBULA_API_KEY=neb_xxx\nNEBULA_COLLECTION_ID=your_collection_id" >> ~/.openclaw/.env && openclaw gateway restart
    ```
  </Tab>

  <Tab title="Config File">
    Edit `~/.openclaw/openclaw.json`:

    ```json theme={null}
    {
      "plugins": {
        "entries": {
          "openclaw-nebula": {
            "enabled": true,
            "config": {
              "apiKey": "your_api_key_here",
              "collectionId": "your_collection_id_here"
            }
          }
        }
      }
    }
    ```

    Then restart: `openclaw gateway restart`
  </Tab>
</Tabs>

### Options

| Option           | Default | Description                              |
| ---------------- | ------- | ---------------------------------------- |
| `collectionName` | --      | Optional display name for the collection |
| `autoCapture`    | `true`  | Auto-store conversations after AI turns  |
| `debug`          | `false` | Enable debug logging                     |

## Usage

Conversations are automatically captured as memories after each AI response. This is enabled by default via `autoCapture`.

### Agent Tool

The AI agent can search stored memories during conversations:

#### `nebula_search`

Search stored memories for relevant context. The agent is encouraged to proactively use this tool to recall user preferences, past discussions, and relevant information before responding.

```typescript theme={null}
nebula_search({ query: "coding preferences" })
```

### Slash Commands

#### `/nebula-remember <text>`

Manually save to memory.

```
/nebula-remember I prefer TypeScript for new projects
```

#### `/nebula-recall <query>`

Search and display memories.

```
/nebula-recall typescript preferences
```

### CLI

Search memories directly from the terminal:

```bash theme={null}
openclaw nebula search "python coding style"
```

## Troubleshooting

### Plugin won't load

**`nebula: apiKey is required`**

Set the `NEBULA_API_KEY` environment variable or add it to your config file.

**`nebula: collectionId is required`**

Set the `NEBULA_COLLECTION_ID` environment variable or add it to your config file. Get your collection ID from [zeroset.com](https://zeroset.com).

### No search results

<Info>Nebula takes 5-10 seconds to index new memories. Wait after storing before searching.</Info>

### Debug mode

Enable verbose logging to diagnose issues:

```json theme={null}
{
  "config": {
    "debug": true
  }
}
```

Check OpenClaw logs for `[nebula]` prefixed messages.