● Model Context Protocol server

Turn agent data into visual stories.

Plotable gives AI agents a safe, structured way to publish interactive charts, tables, and KPI dashboards—then share them through a permanent URL.

Configure your client See how it works

✓ Declarative specs ✓ Versioned artifacts ✓ MCP Apps ready

MCP · publish


                  // Your agent calls one tool
                  {
                    "tool": "publish",
                    "title": "Revenue overview",
                    "data": [ ... ],
                    "spec": { "components": [ ... ] }
                  }

From rows to a live page

A visualization pipeline designed for agents.

No generated HTML and no one-off deployments. Plotable keeps data, presentation, and version history in one predictable workflow.

Send structured data

Provide JSON rows or CSV from any MCP-compatible client, along with a declarative visualization spec.

Validate and publish

Plotable verifies fields and data types, stores the artifact, and renders it using trusted components.

Share and evolve

Use the generated URL anywhere. Updates preserve the same slug while every historical version remains available.

Four focused tools

A small MCP surface that is easy to reason about.

Agents can create, revise, retrieve, and discover artifacts without managing files, frontend code, or deployments.

publish

Create

Turn data and a visualization spec into a new shareable artifact.

update

Revise

Change the data, the spec, or both while preserving the public slug.

get

Inspect

Retrieve the current artifact or a specific immutable version.

list

Discover

Find recently published artifacts and their latest metadata.

Connect in minutes

Configure Plotable as a remote MCP server.

Add the Streamable HTTP endpoint and the bearer token supplied by your administrator to an MCP-compatible client.

Remote endpoint

Plotable uses the MCP Streamable HTTP transport. Keep your token private and send it in the Authorization header on every request.

https://mcp.plotable.talayolabs.com/mcp

MCP client configuration JSON

{
  "mcpServers": {
    "plotable": {
      "type": "http",
      "url": "https://mcp.plotable.talayolabs.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}

Curated building blocks

Useful visualizations without arbitrary code.

The agent describes intent. Plotable owns rendering, safety, responsive layout, and interactive behavior.

▥

Charts

Bar and line charts with categorical or time-based axes and optional grouped series.

KPI cards

Summarize totals, averages, minimums, maximums, or row counts at a glance.

≡

Tables

Present selected fields in a readable, responsive tabular layout.

⌕

Interactive filters

Add select, search, and date-range controls that work across the entire artifact.

↻

Immutable history

Every update creates a versioned snapshot while the main URL always shows the latest.

◇

MCP Apps

Compatible clients can render the visualization inline using a registered UI resource.

Declarative by design

Describe the view, not the implementation.

Safer output

Agents never submit executable HTML or JavaScript.

Actionable validation

Unknown fields and incompatible data types are rejected before publishing.

Consistent rendering

Every artifact uses the same maintained viewer and security policy.

Example visualization spec JSON

{
  "title": "Monthly revenue by region",
  "components": [
    {
      "type": "kpi",
      "field": "revenue",
      "aggregation": "sum"
    },
    {
      "type": "chart",
      "chart": "bar",
      "x": "month",
      "y": "revenue",
      "groupBy": "region"
    }
  ],
  "filters": [
    { "field": "region", "kind": "select" }
  ]
}

Questions

What to know before connecting.

What data formats can Plotable receive?

Plotable accepts arrays of JSON objects or CSV input. CSV values are normalized into strings, numbers, booleans, nulls, and ISO-like date values when possible.

Does every published artifact require a new deployment?

No. Artifacts are data records rendered by the existing viewer. Publishing or updating one is immediate and does not create a new application deployment.

How are updates and historical versions handled?

The update tool creates a new immutable version under the same artifact slug. The main share URL follows the latest version, while version-specific URLs preserve historical output.

Can agents provide custom HTML or JavaScript?

No. Plotable deliberately accepts only structured data and a validated declarative spec. This keeps rendering within a curated, trusted component set.

Where do I get an access token?

The MCP endpoint uses a bearer token managed by the service administrator. Request a Plotable MCP token from your Talayo Labs administrator and keep it out of source control.

Ready to publish

Give your agent a canvas for its data.

Connect the MCP endpoint, provide your bearer token, and ask your agent to publish a visualization from the data already in your conversation.

View configuration Check service health