MCP Server
Your data,
in any AI agent
Connect Claude, ChatGPT, or your own agent to your data. Terno keeps every query secure, accurate, and governed.
- Read-only for databases: only SELECT queries run.
- Controlled writes for apps: admin approval and dry-run.
- Your rules apply: private tables, columns and rows.
- No passwords in the chat: OAuth sign-in, encrypted credentials.
Using Cursor, Claude Code or your own agent? See setup steps ↓
Connect your agent
Add Terno as a custom MCP connector
https://mcp.terno.ai/mcp {
"mcpServers": {
"terno": { "url": "https://mcp.terno.ai/mcp" }
}
}https://mcp.terno.ai/mcp/preview {
"mcpServers": {
"terno-preview": {
"url": "https://mcp.terno.ai/mcp/preview"
}
}
}
How it works
From your data to any agent
Terno
secure query layer
PostgreSQL 
MySQL 
Oracle 
Snowflake 
BigQuery 
Databricks 
Google Ads 
GA4 
Salesforce Ask from
Any AI agent
App connectors work in the Terno Agent and on the preview MCP endpoint today. The main MCP endpoint covers connected databases and warehouses.
AI clients
Use it from the agent you already have
Terno speaks the open Model Context Protocol, so any MCP client can connect with one URL and an OAuth sign-in.
Terno Agent Built into Terno AI, no setup Built in
Claude claude.ai and Claude Desktop Connect ↗
ChatGPT Business & Enterprise workspaces Connect ↗
Claude Code Terminal coding agent 
Cursor AI code editor 
Antigravity Agentic IDE 
Visual Studio Code With GitHub Copilot 
Windsurf AI code editor 
Gemini CLI Terminal agent 
Lovable AI app builder 
Custom agents Any framework that speaks MCP 
Self-hosted Open-source TernoDBI The layer in between
More than a pipe to your data
Giving an AI raw database access is risky and often wrong. Terno checks every query before it runs.
Security
- SQLShield: only read-only SELECT queries run.
- Access rules: hide tables and columns, limit rows by role.
- Encrypted credentials: entered in Terno, never in the chat.
Accuracy
- Real schema: tables, columns, types, and sample rows.
- Plain-English labels: names and descriptions guide the AI.
- Team context: saved notes and a shared org prompt.
Governance
- OAuth consent: standard sign-in, revocable any time.
- Roles: members read; only org admins can write.
- Safe writes: explicit write permission, with dry-run for app changes.
Examples
Ask your agent
Plain questions in Claude, ChatGPT, or any MCP client, answered from your own data.
PostgreSQL in Claude “Which 10 products brought in the most revenue last month?”
Snowflake in ChatGPT “How many new customers signed up each week this quarter?”
BigQuery in Cursor “Compare average order value by country, this year vs last.”
MySQL in Terno Agent “Which of my databases have a customers table, and what columns does it have?”
PostgreSQL in Claude “Add a note explaining why the legacy_users table should be ignored.”
Google Ads in ChatGPT “Which Google Ads campaigns had the lowest cost per conversion?”
The Google Ads example uses the preview endpoint. Every other example works on the main endpoint today.
How to connect
Connect in a minute
Just want answers? The Terno Agent in Terno AI needs no setup. Open Terno AI →
claude.ai and Claude Desktop. Free plans can add one custom connector.
- 1 Open Customize → Connectors.
- 2 Click + Add, then Add custom connector.
- 3 Name it Terno, paste the server URL, and click Continue.
- 4 Sign in to Terno and approve access.
On Team and Enterprise plans, an Owner adds it once under Organization settings → Connectors; members then click Connect.
Copy this
https://mcp.terno.ai/mcpChatGPT Business and Enterprise/Edu on the web. Workspace admins and owners add it.
- 1 Open Plugins, select +, then Add custom MCP server.
- 2 Name it Terno and paste the server URL under Connection.
- 3 Choose OAuth, then acknowledge the risk warning.
- 4 Select Create as a plugin and sign in to Terno.
Copy this
https://mcp.terno.ai/mcpTerminal coding agent.
- 1 Run the command below in your terminal.
- 2 Start Claude Code, run /mcp, and choose terno to sign in.
Copy this
claude mcp add --transport http terno https://mcp.terno.ai/mcpAI code editors.
- 1 Cursor: add the config below to ~/.cursor/mcp.json (or .cursor/mcp.json in a project).
- 2 VS Code: run MCP: Add Server from the Command Palette, choose HTTP, and paste the server URL.
- 3 Sign in to Terno when the editor prompts you.
Config
{
"mcpServers": {
"terno": { "url": "https://mcp.terno.ai/mcp" }
}
}Any framework that speaks MCP, or a self-hosted layer.
- 1 Connect your MCP client library to the server URL over Streamable HTTP.
- 2 Authenticate with OAuth; dynamic client registration is supported.
- 3 Prefer to run it yourself? Self-host the open-source TernoDBI layer.
Config
pip install terno-dbi
ternodbi startFrequently Asked Questions
What is the Terno MCP server? +
It is a hosted Model Context Protocol endpoint at https://mcp.terno.ai/mcp. It lets AI assistants such as Claude and ChatGPT query the data sources you have connected in Terno, through Terno's secure query layer.
Can the AI change or delete my data? +
Not in your databases or warehouses: every SQL query is checked by SQLShield, and only read-only SELECT statements are allowed. Some app connectors, such as Google Ads, can also take actions; these need an organisation admin and write permission, and can be previewed with a dry run first. Write access, available only to organisation admins, changes Terno's own records such as descriptions and notes, never your database.
Which data sources work over MCP? +
Databases and warehouses connected in Terno, such as PostgreSQL, MySQL, Snowflake, BigQuery, Databricks, and Oracle. App connectors such as Google Ads and GA4 work in the Terno Agent today, and over MCP through the preview endpoint https://mcp.terno.ai/mcp/preview. They are coming to the main endpoint soon.
Do I share my database password with the AI? +
No. Connection details are entered directly in Terno over an authenticated web session. The assistant only receives a link to that page, never a credential.
How do I revoke access? +
Remove the connector in your AI app, or revoke it from your Terno account at any time.
Contact Us
Have a question, need a demo, or want to work with us? Reach out and our team will get back to you shortly.
Get in touch
We’re here to help you with data insights, onboarding, pricing or anything else.
reachus@terno.ai
Phone
+1 (412) 568-3901
Address
9450, SW Gemini Dr, PMB 90168, Beaverton, Oregon 97008-7105, United States