Pigment MCP Server

Connect AI assistants to live Pigment business planning data: query FP&A metrics, workforce models, and sales forecasts in natural language.

Analytics by Pigment OAuth2 active

Connect (Remote Server) Docs

Overview

The Pigment MCP Server is an officially hosted integration that connects Pigment's enterprise business planning platform to MCP-compatible AI assistants like Claude, ChatGPT, Mistral, Gemini, Perplexity, ServiceNow, and Agentforce. It exposes live Pigment data, models, metrics, and native Agents through the Model Context Protocol so business users can ask natural-language questions like "What is our revenue by product for the last 6 months?" and receive answers grounded in their workspace's real planning data.

Pigment is widely used for Financial Planning & Analysis (FP&A), Sales Performance Management (SPM), Workforce Planning, and Supply Chain & Operations. The MCP server lets AI agents discover applications, inspect metric structure (dimensions, scenarios, data types), and query AI-enabled metrics under strict role-based access control. Only metrics flagged with "AI Data Access" and using Number, Integer, or Boolean data types are exposed.

The server runs in two modes. Default (read-only) mode includes discovery and query tools. Advanced mode adds a search tool that surfaces full application metadata (block names, formulas, dimensions) and write capabilities for creating and editing Dimensions, Metrics, Views, and Boards. All access respects existing Pigment permissions: users cannot retrieve anything through MCP that they could not see in the Pigment UI.

Tools

Tool	Description
`get_organization_info`	Get information about the current organization (Workspace), including ID and name.
`get_applications`	List all Pigment applications the user can access, with their ID, name, and description.
`get_ai_metrics`	List AI-enabled Metrics in an Application with their IDs and names.
`get_metric_description`	Get the structure of a Metric: name, data type, description, Dimensions (with sample Items), and Scenarios.
`query_data`	Retrieve data from a single Metric using a natural-language query, with optional dimension filtering.
`search (advanced mode)`	Scan and understand the entire Application logic. Exposes all Block metadata (names, data types, Dimensions, formulas) but does not return actual data values.
`write tools (advanced mode)`	Create and edit Dimensions, Metrics, Views, and Boards within a Pigment Application. Specific tool names not enumerated in public docs.

Setup Guide

The Pigment MCP Server is provider-hosted. There is nothing to install locally: Pigment generates a unique MCP endpoint URL per workspace.

Prerequisites

An active Pigment workspace
Pigment AI must be set up in the workspace first
Workspace admin permissions to enable MCP
"AI Data Access" enabled on each Metric you want to query (only Number, Integer, and Boolean data types are supported)

Steps

In Pigment, go to Workspace Settings, Integrations, then the MCP tab.
Toggle the MCP (Model Context Protocol) setting on.
Copy the auto-generated MCP Endpoint URL (unique to your workspace; Test and Deploy environments each need their own).
Add the endpoint to your MCP-compatible client (Claude, ChatGPT, Mistral, etc.).

Claude Desktop / generic MCP client config

{
  "mcpServers": {
    "pigment": {
      "url": "https://<your-workspace>.pigment.app/mcp",
      "transport": "http"
    }
  }
}

Replace the URL with the exact endpoint shown in your Pigment workspace MCP settings. Authentication uses your existing Pigment credentials and respects role-based access policies.

Note: Each workspace and each environment (Test, Deploy) requires its own endpoint URL and client configuration.

Use Cases

Ask "What is our forecasted revenue by product line for Q3?" inside Claude or ChatGPT and get values pulled live from a Pigment FP&A model.
Pull headcount and compensation totals from workforce planning models to brief execs without opening Pigment.
Investigate variance: have the agent list AI-enabled metrics in the Sales Performance app, then query a quota attainment metric filtered by region and rep.
Use advanced-mode search to let an AI map out the structure of a Pigment application (blocks, formulas, dimensions) when onboarding a new analyst.
Trigger Pigment Agent missions, like a demand-forecast refresh, from an external assistant workflow.

Example Prompts

"List the Pigment applications I have access to and summarize what each is for."
"Show me the structure of the Revenue Forecast metric: dimensions, scenarios, and data type."
"Query the Headcount metric for the EMEA region across all departments for the next 4 quarters."
"Compare actual vs forecast sales by product for the last 6 months from our Pigment SPM app."
"Search our FP&A application for any metrics that use the OpEx formula and list their dependencies."

Pros

Officially built and hosted by Pigment, no self-hosting or maintenance required.
Enforces existing Pigment RBAC and audit logging: users can only query data they already have permission to see.
Compatible with most major MCP clients (Claude, ChatGPT, Mistral, Gemini, ServiceNow, Agentforce).
Two modes give a clean read-only default plus an advanced mode for write operations and full application introspection.

Limitations

Requires a paid Pigment workspace and Pigment AI to be enabled; not available to non-customers.
Only Number, Integer, and Boolean metrics are queryable; text and date-typed metrics are excluded.
Admins must explicitly enable "AI Data Access" on each metric, which adds setup overhead in large models.

Alternatives

Anaplan: similar enterprise planning platform, with its own emerging AI integrations.
Cube MCP server: open-source semantic layer that can expose financial and analytics models to AI agents.
Snowflake or BigQuery MCP servers: general-purpose data warehouse access if planning data lives there rather than in a dedicated EPM tool.