Plan availability | All plans |
Permissions |
|
Platform(s) | Works with any AI assistant that supports Model Context Protocol (Claude, ChatGPT, Cursor, and other MCP-compatible tools) |
Stackby's MCP server lets you connect AI tools directly to your stacks and tables. Once connected, you can ask questions about your data, create and update records, and automate workflows—all through natural conversation with AI tools like Claude, ChatGPT, Cursor, and more.
Model Context Protocol (MCP) is an open standard that allows AI tools to securely connect to external data sources and services. With MCP, AI assistants can query your Stackby data, perform actions on your behalf, and present structured information—all while respecting your existing permissions and access controls.
The Stackby MCP Server enables several types of interactions between your AI tools and your data:
Search and analyze data
Ask your AI tool questions about your Stackby data:
"Show me all open tasks in my Project Tracker stack assigned to the engineering team"
"What are the top 5 most requested features in our feedback table?"
"Find all CRM contacts from Mumbai who haven't been followed up with this month"
Add new information to your stacks through conversation:
"Add these 3 new leads to my Sales Pipeline stack"
"Create a new task called 'Update onboarding docs' in the Q2 Sprint table, due Friday"
Make changes to your data without opening Stackby:
"Mark task #47 as Complete in the Sprint table"
"Update the budget for the 'Website Redesign' project to $12,000"
Use AI to extend your stacks with new tables and fields:
"Add a new column called 'Priority' with options Low, Medium, High to the Tasks table"
"Create a new table called 'Client Feedback' in my CRM stack"
Before setting up the Stackby MCP Server, you will need:
A Stackby account (any plan)
A Stackby API key (see setup steps below)
An MCP-enabled AI host such as Claude, ChatGPT, or Cursor
Data already added or stored in at least one Stackby stack
"clientId": "sb_cli_VjQkRhT7_Mx7gY6xi4ro-s-4",
"clientSecret": "sb_sec_5mYL5ndGlswGx4l-91fUAfGZ9FW8YU51"
Stackby's MCP server uses API key authentication. Follow these steps to connect your AI assistant.
Log in to your Stackby account at stackby.com.
Click your profile avatar in the top-right corner and select Account Settings.
Navigate to the API section.
Copy your personal API key.
📝 Note Keep your API key private. Anyone with your API key can read and write data in your stacks on your behalf. |
In your AI assistant's MCP configuration, add a new server entry with the following details:
Stackby authenticates using a custom request header. Add the following header to your MCP server configuration:
x-api-key: YOUR_STACKBY_API_KEY
{
"mcpServers": {
"stackby": {
"command": "npx",
"args": ["-y", "@anthropic-ai/mcp-client-http"],
"env": {
"MCP_SERVER_URL": "https://mcp.stackby.com/mcp",
"MCP_HEADERS": "{\"x-api-key\": \"YOUR_STACKBY_API_KEY\"}"
}
}
}
}Once configured, ask your AI assistant to run a quick check:
"Can you ping the Stackby MCP server and list my workspaces?"
If the connection is working, your AI assistant will return a list of your Stackby workspaces.
Get the most out of your MCP integration with these tips:
Be specific: Mention the stack name or table name when asking about data. The more context you give, the more accurate the results.
Use natural language: You don't need to know field IDs or table IDs—describe what you want in plain English.
Describe your stack first: For complex queries, briefly explain the structure of your stack before asking for data (e.g., "My CRM stack has a Contacts table with Name, Company, Status, and Last Contacted columns").
Check permissions: If you can't access certain data via MCP, verify that your API key belongs to a user who has permission to view or edit that stack.
Think about data access: Your AI tool will be able to read and write your Stackby data on your behalf. Only connect tools you trust.
The Stackby MCP server currently provides the following tools:
Tool name | Description |
list_workspaces | Lists all Stackby workspaces you have access to. Use this to find your workspace ID. |
list_stacks | Lists all stacks (databases) you can access in your Stackby account. Optionally filter by workspaceId. Use this to get the stackId needed for other tools. |
list_tables | Gets all tables in a given stack, including their names and IDs. |
describe_table | Returns the full schema of a table: field names, types, IDs, and available views. Essential before filtering on select/multiSelect fields. |
list_records | Lists rows from a table with support for filtering, sorting, pagination, and view selection. Always obtain stackId and tableId from list_stacks and list_tables first. |
search_records | Searches for rows containing a text string in one or more specified columns. |
get_record | Retrieves a single row by its record ID. |
create_record | Creates a new row in a table. Field values are keyed by column name. |
update_records | Updates up to 10 existing rows per request. Unspecified fields remain unchanged. |
delete_records | Soft-deletes up to 10 rows by ID. |
create_table | Creates a new table inside an existing stack. |
create_field | Adds a new column (field) to a table. Supports types such as shortText, number, checkbox, singleOption, multipleOptions, dateAndTime, email, url, and more. |
ping | Pings the MCP server to verify it is running and reachable. |
Keep these limitations in mind when using the Stackby MCP Server:
Rate limits: MCP server calls are subject to Stackby's standard API rate limits.
Batch size: Creating, updating, and deleting records is limited to 10 records per request.
Soft deletes: Records deleted via MCP are soft-deleted and may be recoverable from within Stackby.
Permission enforcement: The MCP server respects your existing Stackby permissions. You can only access or modify data that your API key's user account has permission to access.
Schema changes: Tool names, behaviors, and available capabilities may change as the MCP server is updated.
Yes. If your Stackby account has access to a stack, the MCP server can read and write data within it, subject to your permission level. Viewer-level access allows reading via MCP but not creating or updating records.
Any AI assistant that supports the Model Context Protocol can connect to Stackby's MCP Server. This includes Claude, ChatGPT, Cursor, and other MCP-compatible tools. Check your AI assistant's documentation to confirm MCP support.
Navigate to Account Settings > API in Stackby and regenerate your API key. This will immediately invalidate your old key and disconnect any AI tools that were using it.
MCP uses the Stackby public API under the hood and will count against existing API rate limits. Records created using MCP count against your plan's row limits.
No. The MCP integration is available with your current Stackby plan at no extra charge, subject to standard API rate limits.
The MCP server will return an error, just as the Stackby API would. Your AI assistant should communicate this limitation to you and suggest alternatives.
Currently, the MCP server supports creating tables and fields within existing stacks, but does not support creating brand-new stacks. Stack creation should be done directly within the Stackby interface.