# flowcore/mcp-flowcore-platform [Health: Active]

**Category:** 📊 Data Platforms  
**Repository:** https://github.com/flowcore-io/mcp-flowcore-platform  
**GitHub Stars:** 9  
**Views:** 4  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/flowcore-mcp-flowcore-platform

## Description
Interact with Flowcore to perform actions, ingest data, and analyse, cross reference and utilise any data in your data cores, or in public data cores; all with human language.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "mcp-flowcore-platform": {
    "command": "npx",
    "args": ["-y","You"],
    "env": {
      "USERNAME": "",
      "PAT": ""
    }
  }
}
```

**Requires environment variables:** `USERNAME`, `PAT` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What flowcore/mcp-flowcore-platform MCP server does

The flowcore/mcp-flowcore-platform MCP server gives MCP-compatible AI assistants a structured way to interact with the Flowcore Platform. Its stated scope includes managing and using Flowcore resources, ingesting data, analysing data, and cross-referencing information held in data cores. The available information also says that agents can work with both the user's data cores and public data cores.

Requests are intended to be expressed in human language. MCP provides the interface between the assistant and the platform, while Flowcore remains the system where the relevant data and resources are managed. The repository does not provide a complete tool-by-tool inventory in the supplied material, so individual operation names and parameter details should be confirmed from the package or source before building tightly coupled workflows.

## How it works

The package starts an MCP server locally and exposes a standardized interface for Flowcore operations. An assistant sends an interpreted request through MCP; the server then uses the supplied Flowcore credentials to access the platform and return the result to the client.

The project supports several launch paths: direct execution with `npx`, a globally installed npm package, the development source through Bun, or the compiled JavaScript output through Node. The documented package name is `@flowcore/platform-mcp-server`.

## Setup and configuration

The documented direct-run command is:

```bash
npx -y @flowcore/platform-mcp-server --username <username> --pat <pat>
```

Replace the placeholders with a Flowcore username and PAT when configuring the command. The README also identifies `USERNAME` and `PAT` as required environment variables, representing the same two credentials. A PAT is a Flowcore Personal Access Token and should be handled as a secret.

For a global installation, install `@flowcore/platform-mcp-server` with npm and start it using the `platform-mcp-server` executable with the username and PAT options. Developers can instead use Bun to install dependencies and run `src/index.ts`, or build the project and execute `dist/cli.js` with Node.

## Tools and capabilities

Based on the supplied description, the server supports these capability areas:

- Perform actions in the Flowcore Platform.
- Ingest data into Flowcore.
- Analyse information stored in data cores.
- Cross-reference data across available cores.
- Use information from private or public data cores.
- Manage and interact with Flowcore resources through MCP.

The material does not specify whether each capability is exposed as a separate MCP tool, resource, or prompt, nor does it document access controls for individual data cores.

## Limitations and notes

The flowcore/mcp-flowcore-platform MCP server requires Flowcore credentials; it is not described as an anonymous connector. The supplied documentation does not list supported MCP clients, provide a detailed API or tool reference, or state the permissions required by the PAT. Validate those details before deploying it in a production agent.

The README points to a separate Flowcore local read-model MCP server for workloads that need reduced hallucinations, lower token use, larger query coverage, or faster querying. That is a different project and is not part of this server's installation or capability set. The repository excerpt also does not state a software license or pricing policy.

_Full upstream README: https://allmcps.com/mcp/flowcore-mcp-flowcore-platform/readme_