Let compatible AI clients understand the active workspace, propose a complete plan, and ask before every reversible action. The demo below shows the full journey from request to plan.
CONNECTION GUIDE
Connect an AI client in three steps.
SmartPlot detects the MCP connector shipped with the app. The page never exposes a personal folder path; copy either a terminal command or a safe AI setup prompt from Settings.
01
Open SmartPlot and your project
Keep the target project in the foreground. MCP only reads and operates on the active workspace.
02
Configure your AI client once
For the first connection, open Settings → AI & MCP and copy either the terminal command or the safe setup prompt for your AI client.
03
Start a new AI task and connect
Submit a plotting request. Once connected, the SmartPlot header displays “MCP connected”.
AFTER FIRST-TIME SETUP
Start a new task in your AI client or reload MCP. You do not need to restart SmartPlot, and previously connected clients can go straight to the next task.
INTERACTIVE DEMOFrom AI chat to a SmartPlot plan
AI
External AI clientConnected to SmartPlot MCP
You
MCP
SmartPlot workflowPlan preview · not executed
PLANNING
01
Read the active data
Check numeric columns, grouping information, and the current workspace.
WAITING
02
Create the scatter plot
Map X and Y while preserving the real data range and point distribution.
WAITING
03
Analyze marginal curves
Add kernel-density estimates along X and Y to show marginal distributions.
WAITING
04
Apply a soft palette
Use a translucent pastel palette with clear visual hierarchy.
WAITING
CONFIRM EACH STEP
This demo only shows plan generation. It does not change any data or figures.
MCP CAPABILITIES
From workspace context to an approvable workflow.
Read-only capabilities return only the privacy-safe structure needed for the task. Controlled actions require foreground approval or authorization for the active workflow. These are product capabilities, not background scripts detached from the interface.
READ
Read the active workspace
Read status, a privacy-safe project summary, table structure, and current figure styles without exposing raw data, file paths, or internal IDs.
PLAN
Plan complete workflows
Organize a plan across data preparation, plotting, analysis, appearance, and composition, then let you choose how to approve it.
DATA
Process data and units
Sort, filter, clean, normalize, calculate columns, fill missing values, remove duplicates, and convert display units or numeric values.
PLOT
Create native figures
Create figures through SmartPlot’s available entry points, including line, scatter, bubble, error-bar, column, dual-Y, heatmap, radar, Venn, volcano, histogram, and box plots.
STYLE
Style color, lines, and markers
Adjust color, opacity, line weight, dash style, markers, and complete palettes. Plans remain editable in the docked SmartPlot panel.
ANALYZE
Use native analysis tools
Use available smoothing, baseline, normalization, trend or peak fitting, integration, derivatives, distribution fitting, significance, and marginal distributions.
ANNOTATE
Annotate, highlight, and layer
Add annotations and range highlights, reorder selected objects, and continue into the figure composer.
MCP
Visible execution, only in SmartPlot
MCP can only operate objects in the active project. Changes appear immediately and enter undo history; switching projects, disconnecting, or quitting SmartPlot ends the workflow authorization.
REAL PRODUCT UI
The plan returns to SmartPlot for approval.
This capture comes from a real MCP connection to the V1.75.1 debug build. The page uses a 2× PNG enhancement and provides a direct full-size view to avoid another lossy encode.
Real interface capture · plan generated, no changes executedOpen original size
V1.75 DEVELOPER PREVIEW
MCP ships with SmartPlot desktop.
Download SmartPlot, open a project, then connect a compatible AI client using the setup shown in Settings.
