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

# Local Workflow

> Build a workflow with Airstore in local mode

<Note>
  This guide covers **local mode** for self-hosted deployments and development. For the standard cloud experience, see the [Quickstart](/quickstart).
</Note>

This guide shows how to use Airstore locally with your own MCP servers. You'll configure tools, mount them as a filesystem, and use them with Claude Code.

## Prerequisites

* [Airstore CLI installed](/installation)
* Node.js (for npx-based MCP servers)
* FUSE dependencies (installed by the install script)

## Step 1: Configure your tools

Create a `config.local.yaml` file:

```yaml theme={null}
mode: local

gateway:
  grpc:
    port: 1993
  http:
    host: 127.0.0.1
    port: 1994

tools:
  mcp:
    filesystem:
      command: npx
      args: ["-y", "@modelcontextprotocol/server-filesystem", "/tmp", "/Users"]

    memory:
      command: npx
      args: ["-y", "@modelcontextprotocol/server-memory"]

    wikipedia:
      command: npx
      args: ["-y", "@modelcontextprotocol/server-wikipedia"]
```

## Step 2: Mount the filesystem

```bash theme={null}
airstore mount ~/airstore --config config.local.yaml
```

Your tools are now available:

```bash theme={null}
ls ~/airstore/tools/
# filesystem  memory  wikipedia
```

## Step 3: Use the tools

Each tool works like a regular CLI command:

```bash theme={null}
# Search Wikipedia
~/airstore/tools/wikipedia search "artificial intelligence"

# List files
~/airstore/tools/filesystem list_directory /tmp

# Store information in memory
~/airstore/tools/memory create_entities '[{"name": "project", "type": "note"}]'
```

## Step 4: Run Claude Code

```bash theme={null}
cd ~/airstore
claude "Use the wikipedia tool to research Alan Turing and summarize his contributions"
```

Claude Code can invoke the tools directly and combine results.

## Why local mode?

| Use case     | Description                                     |
| ------------ | ----------------------------------------------- |
| Development  | Test MCP servers before deploying to cloud      |
| Self-hosted  | Run everything on your own infrastructure       |
| Custom tools | Use internal MCP servers not available publicly |
| Offline      | Work without internet connectivity              |

## Chaining tools

Because tools output JSON, you can chain them with standard bash:

```bash theme={null}
# Search Wikipedia and filter results
~/airstore/tools/wikipedia search "physics" | jq '.results[:3]'

# Store search results in memory
RESULTS=$(~/airstore/tools/wikipedia search "Alan Turing")
~/airstore/tools/memory create_entities "[{\"name\": \"turing-research\", \"type\": \"note\", \"observations\": [\"$RESULTS\"]}]"
```

## Adding more tools

Add tools to your config and remount:

```yaml theme={null}
tools:
  mcp:
    # ... existing tools ...
    
    github:
      command: npx
      args: ["-y", "@modelcontextprotocol/server-github"]
      env:
        GITHUB_TOKEN: "${GITHUB_TOKEN}"
```

```bash theme={null}
# Unmount (Ctrl+C) and remount
export GITHUB_TOKEN=ghp_xxxxxxxxxxxx
airstore mount ~/airstore --config config.local.yaml
```

Now you have GitHub access too:

```bash theme={null}
~/airstore/tools/github list_repositories --owner=acme
```

## Next steps

<CardGroup cols={2}>
  <Card title="Tools concept" icon="wrench" href="/concepts/tools">
    Learn more about how tools work.
  </Card>

  <Card title="Config reference" icon="file-code" href="/reference/config">
    Full configuration options.
  </Card>
</CardGroup>