Model Context Protocol (MCP)
Altertable supports the Model Context Protocol (MCP), an open standard that lets AI agents connect to external systems through a shared tool interface.
Through MCP, AI assistants can connect directly to Altertable to inspect your environment, query data, review existing assets, and trigger actions without leaving the conversation.
Learn more: Read about AI Agents to understand how Altertable agents work with MCP.
What Agents Can Do
When an AI agent connects to Altertable via MCP, it can:
- Inspect the current workspace: Discover which organization and environment the session is authorized for
- Explore catalogs and models: List all catalogs in your lakehouse - both Altertable and external catalogs - inspect their schemas, and work with semantic models
- Query the lakehouse with SQL: Run read-only SQL queries, validate them, and inspect query plans in the analytical database
- Review existing assets: Open saved insights, dashboards, memories, organization knowledge, and documentation
- Preview analyses before saving: Test SQL, semantic, funnel, retention, and segmentation insights
- Open interactive viewers: In clients that support MCP apps/resources, render insight previews and SQL query results in Altertable-provided viewers instead of reading raw tool payloads
- Create new outputs: Save insights and memories when write access is needed
- Automate follow-ups: Create scheduled tasks for recurring monitoring and review work
- Use agent skills and context: Search built-in skills for the right workflow, search memories and docs, and capture screenshots of shared dashboards when you need visual context
For the exact tool schema, see Available Tools. For example, the MCP create_task tool accepts a cron_expression when you want the client to schedule recurring work directly.
Getting Started
Follow the Connecting Clients guide to set up your MCP client, authorize with OAuth, and start prompting. For technical details on the OAuth flow and discovery endpoints, see Authentication.
If you are new to Altertable, start with Insights, Dashboards, and the Analytical Database so the MCP tool surface is easier to interpret.
Learn More
- Connecting Clients: End-to-end setup flow for MCP-compatible clients
- Authentication: Endpoint, OAuth discovery, dynamic client registration, and environment scoping
- AI Agents: Understand how agents support MCP-powered workflows
- Analytical Database Client Interfaces: Other ways to access Altertable data outside MCP
- MCP Official Documentation
- MCP Specification