actual-budget-mcp
MCP server for Actual Budget that enables AI agents to read accounts, transactions, categories, and budgets, as well as create and update transactions and transfers.
README
Actual Budget MCP Server
An MCP server for Actual Budget built with FastMCP and actualpy.
Requirements
- Python 3.11+
- uv
- A running Actual Budget server
Setup
git clone https://github.com/Saikumarmohan/actual-budget-mcp
cd actual-budget-mcp
uv sync
Configuration
All configuration is done via environment variables. Pass them via your MCP client config.
| Variable | Required | Default | Description |
|---|---|---|---|
ACTUAL_BASE_URL |
✅ | - | URL of your Actual Budget server |
ACTUAL_PASSWORD |
✅ | - | Actual Budget server password |
ACTUAL_FILE |
✅ | - | Budget file name or sync id |
ACTUAL_CURRENCY |
❌ | USD |
Currency code e.g. INR, EUR |
ACTUAL_CURRENCY_SYMBOL |
❌ | $ |
Currency symbol e.g. ₹, € |
ACTUAL_TIMEZONE |
❌ | UTC |
Timezone e.g. Asia/Kolkata, America/New_York |
ACTUAL_MCP_TRANSPORT |
❌ | stdio |
Transport mode — stdio for Cline/Cursor, http for Docker |
ACTUAL_MCP_HOST |
❌ | 0.0.0.0 |
Host to bind to in HTTP mode |
ACTUAL_MCP_PORT |
❌ | 8000 |
Port to bind to in HTTP mode |
MCP Client Config (Cline / Cursor)
Replace /path/to/actualbudgetpy with the absolute path to this project on your machine.
{
"mcpServers": {
"actualbudget": {
"type": "stdio",
"command": "uv",
"args": [
"run",
"--directory",
"/path/to/actualbudgetpy",
"-m",
"actualbudgetpy.server"
],
"env": {
"ACTUAL_BASE_URL": "http://localhost:5006",
"ACTUAL_PASSWORD": "your_password",
"ACTUAL_FILE": "your_file_id_or_name",
"ACTUAL_CURRENCY": "USD",
"ACTUAL_CURRENCY_SYMBOL": "$",
"ACTUAL_TIMEZONE": "UTC"
}
}
}
}
Note: On Windows, use the full path to
uv.exeinstead of justuve.g.C:\\Users\\username\\.local\\bin\\uv.exe
HTTP Mode (Local)
For running as a standalone HTTP server locally:
- Copy
.env.exampleto.envand fill in your values:
cp .env.example .env
- Set transport to
httpin your.env:
ACTUAL_MCP_TRANSPORT=http
- Start the server:
uv run -m actualbudgetpy.server
- Point your MCP client to
http://localhost:8000/mcp:
{
"mcpServers": {
"actualbudget": {
"type": "streamable-http",
"url": "http://localhost:8000/mcp"
}
}
}
Docker
# build
docker build -t actualbudgetpy .
# run
docker run -p 8000:8000 \
-e ACTUAL_BASE_URL=http://your-server:5006 \
-e ACTUAL_PASSWORD=your_password \
-e ACTUAL_FILE=your_file_id \
-e ACTUAL_CURRENCY=USD \
-e ACTUAL_CURRENCY_SYMBOL=$ \
-e ACTUAL_TIMEZONE=UTC \
actualbudgetpy
Then point your MCP client to http://localhost:8000/mcp.
Available Tools
Read
| Tool | Description |
|---|---|
list_accounts |
List all accounts |
list_payees |
List all payees |
list_transactions |
List transactions with optional filters — date, account, category, payee, notes |
list_categories |
List all categories, filterable by income or expense |
list_category_groups |
List all category groups |
list_budgets |
List budget allocations by month |
Write
| Tool | Description |
|---|---|
add_transaction |
Create a new transaction |
update_transaction |
Update an existing transaction |
delete_transaction |
Delete a transaction by id |
create_transfer |
Transfer money between two accounts |
set_budget |
Set budget amount for a category in a month |
Resources
Resources are read-only data the AI can attach as context at the start of a conversation.
| Resource | Description |
|---|---|
budget://accounts |
Live list of all accounts |
budget://categories |
Live list of all categories |
Prompts
Reusable prompt templates that guide the AI through common workflows.
| Prompt | Parameters | Description |
|---|---|---|
monthly_review |
month (YYYY-MM) |
Full budget vs spending analysis for a month |
spending_by_category |
category, start_date, end_date |
Spending breakdown for a specific category |
add_expense |
description |
Natural language guided expense entry |
Author
Built by Saikumar — feel free to connect!
If you find this useful, consider giving it a ⭐ on GitHub.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。