SqlAugur

SqlAugur provides AI-assisted, read-only access to SQL Server databases. It parses every incoming query into a full SQL syntax tree using Microsoft's T-SQL parser, preventing common injection tricks and ensuring queries are validated at the syntax level before execution. The server exposes a suite of DBA-friendly tooling through a progressive-discovery interface, including schema overviews, table documentation, ER diagrams, and diagnostic helpers. You can use it to explore database schemas, generate documentation, reason about queries, and diagnose performance characteristics, all while enforcing read-only access to protect your data.

To use SqlAugur with an MCP client, configure the MCP server entry with the command name sqlaugur (as shown in the Quick Start). Your client can then send standard MCP commands and leverage the built-in tools such as schema exploration, dependency graphs, query plan analysis, and optional DBA diagnostics integrations. The server exposes a curated set of tools in toolsets, loading additional capabilities on demand to minimize context usage while still enabling powerful insight when needed.

Prerequisites

• .NET 10.0 runtime
• Access to install the SqlAugur global tool (via NuGet) or Docker image as described below

Option 1 — NuGet Global Tool (recommended)

1. Install the global tool:

dotnet tool install -g SqlAugur

2. Create or edit your MCP client configuration to point to the tool:

{
  "mcpServers": {
    "sqlserver": {
      "command": "sqlaugur"
    }
  }
}

3. Run the tool as needed; it will connect to your configured SQL Server instances using the appsettings.json file you provide.

Option 2 — Docker / Podman

1. Run with a mounted config file (example):

docker run -i --rm \
  -v /path/to/appsettings.json:/app/appsettings.json:ro,Z \
  ghcr.io/mbentham/sqlaugur:latest

2. Or pass configuration via environment variables (no config file needed):

docker run -i --rm \
  -e SqlAugur__Servers__production__ConnectionString="Server=host.docker.internal;Database=master;" \
  ghcr.io/mbentham/sqlaugur:latest

3. MCP client configuration example:

{
  "mcpServers": {
    "sqlserver": {
      "command": "docker",
      "args": ["run", "-i", "--rm",
        "-v", "/path/to/appsettings.json:/app/appsettings.json:ro,Z",
        "ghcr.io/mbentham/sqlaugur:latest"]
    }
  }
}

Additional installation notes, including alternative Docker Compose setups and environment-variable approaches, are provided in the official README.

Tips and notes:

• The server is read-only by design: only SELECT and CTE queries are permitted. This reduces risk when exposing the server to AI tools.
• If using Docker, remember to map or pass your SQL Server connection details securely, and consider host networking or host.docker.internal for connecting to a SQL Server running on the host.
• When using the Docker image, you can override the connection string via the SqlAugur__Servers__<env>__ConnectionString environment variable (underscore-delimited) to avoid a config file.
• The MCP client configuration often uses command: "docker" with carefully composed args to run the image; for local tooling, the sqlaugur CLI is typically invoked directly.
• For large deployments, leverage the progressive toolset loading to minimize initial context usage and enable on-demand discovery of additional capabilities.
• If you need to update the tool, use: dotnet tool update -g SqlAugur