Basedash lets Claude Code and Cursor create and edit its charts and dashboards over MCP

A coding agent finishing a piece of analytics work today can produce the query, but the chart or dashboard on top still needs a person to open a dashboard tool and build it. Basedash announced today that its Model Context Protocol server now takes writes: connect Claude Code, Cursor or any MCP client, describe what you want, and the server creates or edits a real Basedash chart or dashboard and returns a link and a rendered screenshot.

The company posted the change to its changelog on 25 September 2026 and launched it on Product Hunt the same day. The entry is titled "Introducing Basedash MCP write: build charts from anywhere". Basedash says every created or updated artifact "comes back with a durable Basedash link and chart screenshots, so you can jump straight from your AI tool into the real thing", and that the new write tools sit alongside "new read tools for listing and fetching charts and dashboards".

What is actually new

Until this week, the Basedash MCP server was read-only. An AI client could ask about data and get answers. It could not build the chart that displayed those answers. That is the shift: Basedash says clients "can create and edit charts and dashboards directly in your workspace using plain-language instructions", and existing MCP connections pick up the new capabilities without a re-authorisation step.

Basedash also says the write path respects existing access controls in the workspace, so the agent inherits whatever the caller is allowed to see or change.

How the round trip works

Basedash's account of the flow is that the client sends a plain-language request ("chart weekly signups, last 90 days"), the server picks the tables, writes the query and the visualisation, saves the artifact in Basedash and returns the link plus an image. The agent's next step in the IDE can quote the link back to the reviewer, or feed the screenshot into the next prompt. Basedash frames this on its Product Hunt page as skipping "schema exploration, SQL validation, and visualisation selection", so "the agent never has to guess at your tables".

The three clients named in the Basedash changelog and Product Hunt page are Claude Code and Cursor, plus "any MCP client". No specific tool names or counts are given in the changelog itself; the Product Hunt page describes the write tools as covering chart creation, chart editing, dashboard creation and dashboard editing.

Basedash is a paid product, not open source

The company positions Basedash as an AI-native business intelligence tool, so the MCP server is a way for an agent to reach data and produce artifacts inside a workspace that already has data connections, access controls and a dashboard model behind it. This is different from an open-source database MCP server that only issues SQL. The write capability lives inside the same product, which is what lets it hand back a durable link to a saved chart rather than a one-off image.

Every claim above is Basedash's own, from the changelog and the product page, and the write path has not been independently tested. Anyone considering it should treat the "any MCP client" promise as needing verification against their own client of choice.

A dashboard that an agent can build and edit changes the shape of an agent workflow that produces reports. The agent finishes its work, files the number, and links to the chart it wrote. What used to be a hand-off is now one more step in the same session.

Source

Announcement: Basedash changelog, "Introducing Basedash MCP write: build charts from anywhere".

Launch: Basedash MCP write on Product Hunt, 25 September 2026.