Quick Start

Deploy an agent, connect a database, and run your first AI-powered query in under 5 minutes.

Prerequisites

A conexor.io account — register free at conexor.io/register
Docker installed on the machine where your database is accessible
A database with a valid connection string (SQL Server, PostgreSQL, or MySQL)

Create a agent in the dashboard

Log in and go to Agents → New Agent. Give it a name (e.g. prod-db) and copy the generated API key — you won't see it again.

NOTEThe agent API key grants access to your MCP server. Store it in a secrets manager, not in plain text.

Deploy the agent with Docker

Run the agent container on any machine that can reach your database. Replace the placeholder values with your agent name and API key from step 1.

bash

docker run -d \
  --name conexor-agent \
  --restart unless-stopped \
  -e PLUGIN_NAME="prod-db" \
  -e API_KEY="your-agent-api-key" \
  -e CLOUD_HUB_URL="https://app.conexor.io/hubs/plugin" \
  conexor/agent:latest

The agent connects to the cloud control plane over WSS (WebSocket Secure). Your database is never contacted by conexor.io — all queries are proxied through the agent.

TIPPrefer bare metal or a Windows service? See the full install guide.

Add a data source

In the dashboard, open your agent and click Add Data Source. Paste your connection string. It is encrypted immediately with AES-256 before leaving your browser — conexor.io never sees it in plaintext.

Connection string formats

PostgreSQLHost=localhost;Port=5432;Database=mydb;Username=user;Password=pass

SQL ServerServer=localhost;Database=mydb;User Id=sa;Password=pass

MySQLServer=localhost;Port=3306;Database=mydb;Uid=root;Pwd=pass

Discover your schema

Click Sync Schema on your data source. conexor.io walks your tables, columns, and foreign keys, then auto-generates a typed MCP tool for each query pattern. This typically takes 5–30 seconds depending on database size.

INFOSchema discovery only reads metadata — it never reads row data. Tables, column names, and data types are fetched from the information schema.

Run your first AI query

Go to MCP Servers, copy your server URL and create an API key. Then point Claude (or any MCP-compatible model) at your server:

python

import anthropic

client = anthropic.Anthropic()

response = client.messages.create(
    model="claude-opus-4-5",
    max_tokens=1024,
    mcp_servers=[{
        "type": "url",
        "url": "https://app.conexor.io/mcp/your-org/your-server",
        "authorization_token": "YOUR_MCP_API_KEY"
    }],
    messages=[{
        "role": "user",
        "content": "How many users signed up in the last 30 days?"
    }]
)

print(response.content[0].text)

Claude will call the appropriate MCP tool, which triggers a parameterized SELECT query through your agent. The result comes back as structured data — never exposing your connection credentials.

Next steps

Install the Agent

More deployment options including bare metal and Windows.

How Plugins Work

Understand the SignalR protocol and agent architecture.

Parameterized Queries

How SQL injection is prevented at the protocol layer.

Audit Logging

Every query logged — how to read and export logs.

All systems operational