Universal Firefly III AI Bridge
Provides comprehensive AI-agnostic access to all Firefly III personal finance features via 66 tools, enabling natural language management of accounts, transactions, budgets, and more.
README
Universal Firefly III AI Bridge (v3.0.0)
A professional-grade, AI-agnostic bridge providing exhaustive 100% API coverage for connecting AI assistants to your Firefly III personal finance instance.
66 tools covering every major Firefly III capability: accounts, transactions, budgets, bills, recurring rules, automation, insights, attachments, currencies, and more.
Compatibility
| AI Platform | Protocol | Connection | Setup Guide |
|---|---|---|---|
| Claude Code | MCP (Native) | stdio | CLAUDE.md |
| Claude Desktop | MCP (Native) | stdio | CLAUDE.md |
| Gemini CLI | MCP Extension | stdio | gemini.md |
| Cursor / VS Code | MCP | stdio or SSE | Manual setup below |
| ChatGPT | OpenAPI Actions | REST / JSON | /openapi.json endpoint |
| Custom Apps | REST API | HTTP | /api/<tool> endpoints |
Installation
Prerequisites
- Node.js v18 or later
- A running Firefly III instance
- A Personal Access Token from Firefly III
(Profile → OAuth → Personal Access Tokens → Create new token)
Clone and install
git clone https://github.com/fabianonetto/mcp-server-firefly-iii.git
cd mcp-server-firefly-iii
npm install
Setup by Platform
Claude Code (Recommended)
Claude Code uses .mcp.json in the project directory for MCP server configuration. Credentials go in a separate .env file so no secrets are ever in config files.
1. Create .env in the repo root (gitignored — never committed):
FIREFLY_URL=http://your-host:PORT
FIREFLY_TOKEN=your_personal_access_token
Get your token: Firefly III → Profile → OAuth → Personal Access Tokens → Create new token
2. Create .mcp.json in the repo root (gitignored — never committed):
{
"mcpServers": {
"firefly-iii": {
"command": "node",
"args": ["./index.js"]
}
}
}
The server reads credentials from .env automatically. No secrets in .mcp.json.
3. Start Claude Code from the repo directory:
claude
The server starts automatically. Claude Code will prompt you to approve it on first launch (once only).
4. Verify the connection by asking:
Use the get_about tool
You should receive your Firefly III version and API information.
Using from other projects:
Copy both .env and .mcp.json to any other project directory, changing ./index.js to the absolute path:
"args": ["/absolute/path/to/mcp-server-firefly-iii/index.js"]
See CLAUDE.md for the full guide including auto-approval configuration.
Claude Desktop
Add the server to your Claude Desktop config file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"firefly-iii": {
"command": "node",
"args": ["/absolute/path/to/mcp-server-firefly-iii/index.js"],
"env": {
"FIREFLY_URL": "http://your-host:PORT",
"FIREFLY_TOKEN": "your_personal_access_token"
}
}
}
}
Restart Claude Desktop after saving.
Gemini CLI Extension (One-command install)
gemini extensions install https://github.com/fabianonetto/mcp-server-firefly-iii
Then configure your instance:
gemini config set extensions.firefly-iii-universal-bridge.settings.FIREFLY_URL "http://your-host:PORT"
gemini config set extensions.firefly-iii-universal-bridge.settings.FIREFLY_TOKEN "your_token"
See gemini.md for the full guide.
Docker (Official Image)
The official image is available on GitHub Packages: ghcr.io/fabianonetto/mcp-server-firefly-iii.
Run as a service (SSE Mode)
Ideal for ChatGPT Actions, Cursor (SSE), or custom integrations.
docker run -d \
--name firefly-mcp \
-p 3001:3001 \
-e FIREFLY_URL="http://your-firefly-instance" \
-e FIREFLY_TOKEN="your_personal_access_token" \
-e PORT=3001 \
ghcr.io/fabianonetto/mcp-server-firefly-iii:latest
Run with Claude Desktop (stdio Mode)
Add this to your claude_desktop_config.json:
{
"mcpServers": {
"firefly-iii": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e", "FIREFLY_URL=http://your-host:PORT",
"-e", "FIREFLY_TOKEN=your_token",
"ghcr.io/fabianonetto/mcp-server-firefly-iii:latest"
]
}
}
}
Cursor / VS Code (MCP Extension)
Add to your MCP config (.cursor/mcp.json or equivalent):
{
"mcpServers": {
"firefly-iii": {
"command": "node",
"args": ["/absolute/path/to/mcp-server-firefly-iii/index.js"],
"env": {
"FIREFLY_URL": "http://your-host:PORT",
"FIREFLY_TOKEN": "your_personal_access_token"
}
}
}
}
HTTP / SSE Mode (ChatGPT Actions, Custom Apps)
Start the server with a port to enable the REST and SSE endpoints:
FIREFLY_URL=http://your-host:PORT FIREFLY_TOKEN=your_token PORT=3000 node index.js
Available endpoints:
| Endpoint | Description |
|---|---|
GET /sse |
SSE transport for MCP clients |
POST /messages |
MCP message handler |
POST /api/<tool_name> |
Direct REST call to any tool |
GET /openapi.json |
OpenAPI 3.0 spec (import into ChatGPT Actions) |
Tool Categories
| Category | Tools | Description |
|---|---|---|
| Core | 1 | System info & connectivity |
| Accounts | 5 | Full CRUD for all account types |
| Transactions | 7 | CRUD, split transactions, search |
| Budgets | 8 | Budgets + monetary limits |
| Bills & Piggy Banks | 7 | Bill tracking + savings goals |
| Automation | 11 | Rules, rule groups, webhooks |
| Recurring | 5 | Recurring transaction rules |
| System | 8 | Currencies + user preferences |
| Insights | 7 | Attachments, charts, net worth, spending |
| Meta | 4 | Categories + tags |
| Object Groups | 2 | Account/piggy bank organization |
| Admin | 1 | Data export |
| Total | 66 |
Documentation
| Document | Description |
|---|---|
| docs/API.md | Complete reference for all 66 tools and their input schemas |
| docs/PROMPTS.md | Prompt examples for common financial tasks |
| docs/USE_CASES.md | Strategic guides: tax assistant, subscription auditor, receipt manager |
| docs/TESTING.md | Test suite documentation (78 tests, all tools covered) |
| CLAUDE.md | Claude Code & Claude Desktop setup guide |
| gemini.md | Gemini CLI extension guide |
Running Tests
npm test
78 tests covering all 66 tools. No live Firefly III instance required — all API calls are mocked. See docs/TESTING.md for details.
Security
- Use a VPN or SSH tunnel if exposing the server to the internet.
- Keep your
FIREFLY_TOKENsecret. Never commit.mcp.jsonor.envfiles. - See SECURITY.md for the full security policy.
Roadmap
- [x] v1.x — Initial connectivity
- [x] v2.x — Exhaustive API coverage (CRUD & core admin)
- [x] v3.x — Power user features (splits, insights, automation)
- [x] v3.1.0 — Official Docker image & CI/CD
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.