New · Streamable HTTP MCP endpoint →

The workspace your AI agents can build in.

Basemind brings databases, applications, dashboards and automations into one place — and exposes all of it over MCP, so Claude, ChatGPT, Codex or Cursor can design schemas, fill rows and ship reports for you.

Orders
Customer portal
Revenue
Daily digest
Agent activity
CustomerPlanStatusMRR
Northwind LabsTeamActive$1,240
Harbor & CoStarterActive$290
Kestrel StudioTeamTrial$0
Oakline FreightBusinessActive$3,860
Juniper HealthStarterTrial$0
Build a revenue dashboard from Orders and publish it to the team.
↳ get_table_schema Orders
↳ create_dashboard_widget MRR by plan
↳ publish_report Revenue
Done. “Revenue” is live with 3 widgets. Every change is logged in Activity and can be undone.
Works with the agents you already use
  • Claude
  • Claude Code
  • ChatGPT
  • Codex
  • Cursor
  • Windsurf
  • Grok
Product

One workspace for data, apps and the work in between.

Everything you build in Basemind is also a tool your agent can call. Click through it yourself, or describe it and let the agent do the clicking.

Databases

Relational tables with grid, kanban, calendar and form views. Describe a schema and it’s created in one step.

Applications

Build portals and internal tools on top of your data, then publish them to your own subdomain.

Dashboards & reports

Charts and summaries that stay live. Publish reports with the access mode you choose, or keep them in the team.

Automations

Triggers and actions that run on row changes, schedules or webhooks — no glue code needed.

Built-in MCP server

Tools for tables, rows, pages, workflows and reports, exposed through one endpoint per workspace.

Project memory

Agents remember decisions and context per project, so the next session picks up where the last one stopped.

Connect MCP

Add Basemind to your agent in two minutes.

One endpoint URL is all a client needs. It works with any tool that supports remote MCP servers.

  1. Create an endpoint

    In Basemind, open Settings → MCP server, click “Create endpoint” and pick the workspace the agent may work in.

  2. Copy the endpoint URL

    Click the copy icon next to the URL. It looks like this: https://basemind.pfgames.io/mcp/<key>/mcp

  3. Add it to your client

    Pick your client and paste the URL where the snippet shows YOUR_ENDPOINT_URL.

Treat the URL like a password. Anyone who has it can act in that workspace. Delete the endpoint in Settings to revoke it at any time.

Run this in your terminal. On claude.ai, add the same URL under Settings → Connectors → Add custom connector.

claude mcp add --scope user --transport http basemind YOUR_ENDPOINT_URL
Safety

Agents move fast. You stay in control.

Every change an agent makes goes through the same permissions and history as yours.

Review before it runs

By default, agent changes wait in the project’s Activity tab for your approval. Turn on Full access per endpoint when you trust the flow.

Logged and undoable

Every agent action is recorded in the project’s Activity tab and can be undone like any other change.

Safe mode projects

Mark a project as Safe mode and agents can read it, but never change it — even with full access.

Give your agent a real place to work.

Create a workspace, connect your MCP client and ask for your first table.

Docs© 2026 Basemind