# gofireflyio/firefly-mcp [Health: Active]

**Category:** ☁️ Cloud Platforms  
**Repository:** https://github.com/gofireflyio/firefly-mcp  
**GitHub Stars:** 16  
**npm Downloads (last month):** 73  
**Views:** 2  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/gofireflyio-firefly-mcp

## Description
Integrates, discovers, manages, and codifies cloud resources with Firefly.

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

```json
"mcpServers": {
  "firefly-mcp": {
    "command": "npx",
    "args": ["-y","@fireflyai/firefly-mcp"],
    "env": {
      "FIREFLY_ACCESS_KEY": "",
      "FIREFLY_SECRET_KEY": ""
    }
  }
}
```

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

## Documentation

## What gofireflyio/firefly-mcp MCP server does

The gofireflyio/firefly-mcp MCP server exposes Firefly capabilities to MCP-compatible AI clients. It works with cloud and SaaS accounts connected to Firefly, allowing an agent to find resources, manage them, and turn discovered resources into Infrastructure as Code.

A typical request might identify an EC2 instance in a specific AWS account and ask the agent to codify it. The server is therefore suited to inventory queries, resource-focused investigation, and creating infrastructure definitions from existing resources.

## How it works

The server is implemented in TypeScript and communicates with the Firefly platform using a Firefly account's access credentials. Clients send natural-language requests through MCP; the server uses those requests to query or act on resources available through the connected Firefly accounts.

The gofireflyio/firefly-mcp MCP server can be launched through NPX. The documented client configuration uses a local command for stdio connections. The README also demonstrates starting the process with an SSE option on port 6001 and connecting through a local `/sse` URL.

## Setup and configuration

Node.js 20 or newer and npm or Yarn are prerequisites. A Firefly account with generated access keys is also required. Install is not necessary for the documented NPX workflow:

```bash
npx @fireflyai/firefly-mcp
```

Provide credentials with `FIREFLY_ACCESS_KEY` and `FIREFLY_SECRET_KEY`, or pass them as `--access-key` and `--secret-key` arguments. For a stdio client configuration, add the NPX command and its environment variables to `mcp.json`. For the SSE example, start the server with `--sse --port 6001`, then configure the client with `http://localhost:6001/sse`.

The gofireflyio/firefly-mcp MCP server is documented for use with Claude and Cursor. Cursor users start the process, connect to it through Cursor's MCP settings, and can then query resources using natural language.

## Tools and capabilities

Documented capabilities include:

- Discovering resources in connected cloud accounts
- Discovering resources in connected SaaS accounts
- Managing resources exposed through Firefly
- Converting discovered resources into Infrastructure as Code
- Authenticating requests with Firefly access and secret keys

## Limitations and notes

The server depends on a Firefly account and valid generated credentials. It requires Node.js 20 or later for the documented setup. Its resource coverage depends on the cloud and SaaS accounts connected to Firefly. Support and configuration details are available through Firefly's documentation or the repository issue tracker.

_Full upstream README: https://allmcps.com/mcp/gofireflyio-firefly-mcp/readme_