Buchhaltungsbutler MCP

Buchhaltungsbutler MCP

Enables AI assistants to manage BuchhaltungsButler bookkeeping through all 48 API endpoints, with safety-categorized tools for read, write, and destructive operations.

Category
访问服务器

README

ohneben's Buchhaltungsbutler MCP

CI Publish Docker image License: MIT

Manage your BuchhaltungsButler bookkeeping in plain language from AI assistants like Claude, Cursor, and any other MCP client.

This Model Context Protocol server exposes the BuchhaltungsButler API v1 — all 48 endpoints, auto-generated from the official OpenAPI spec into MCP tools. Every tool is safety-categorized (read-only / write / destructive) so your assistant knows what an action does before it calls it. Runs over stdio (Claude Desktop and other local launchers) or Streamable HTTP (hosted in Docker).

Why you'll want this

Some MCP servers just forward an API. This one is built to be safe to hand to an LLM and easy to run for real:

What you get Why it matters
All 48 endpoints, auto-generated from the official spec Full coverage of receipts, transactions, postings, invoices and master data — nothing hand-picked or left behind.
Every tool is safety-categorized 🟢 / 🟡 / 🔴 A banner at the top of each tool description tells the model exactly what it does — fetch, create, update, revert or delete — before it acts.
Machine-readable MCP annotations (readOnlyHint, destructiveHint) Hosts that honor annotations (Claude included) can auto-trust reads and demand confirmation before anything destructive.
Two transports: stdio and Streamable HTTP Use it locally in Claude Desktop, or run one always-on server that any number of MCP clients reach over HTTP.
Docker + docker-compose, health check, auto-restart Production-style deployment out of the box: docker compose up and it stays up.
Optional bearer-token auth on the HTTP endpoint Put the server behind a shared secret the moment it's reachable beyond localhost.
Built-in rate limiting Self-throttles under BuchhaltungsButler's 100 requests/customer/minute cap so you never trip it.
Your secrets never reach the model Credentials live in the server's environment and are injected on every request — the assistant only ever sees tool inputs and API responses.

How it compares

At the time of writing this appears to be the only dedicated BuchhaltungsButler MCP server. You could instead point a generic OpenAPI→MCP wrapper at the spec — here's what that leaves on the table:

Capability This project Generic OpenAPI→MCP wrapper*
All 48 BuchhaltungsButler endpoints as tools ✅ ✅
Per-tool 🟢 / 🟡 / 🔴 safety category + banner ✅ ❌
readOnlyHint / destructiveHint MCP annotations ✅ ➖
$ref batch-payload resolution + HTML-stripped descriptions ✅ ➖
Built-in rate limiting (stays under BB's 100/customer/min) ✅ ❌
stdio transport ✅ ✅
Streamable-HTTP transport ✅ ➖
Docker + docker-compose, health check, auto-restart ✅ ❌
Optional bearer-token auth on the endpoint ✅ ❌
Credentials injected server-side, never sent to the model ✅ ➖
License MIT varies

<sub>*Generic OpenAPI→MCP wrappers turn any Swagger/OpenAPI spec into MCP tools. They can reach the same endpoints, but treat every operation identically — no safety categories, no deployment story, and no guardrails tuned for live accounting data. "➖" = varies by tool / not guaranteed. Snapshot from July 2026.</sub>

What you can do

Once it's connected, ask your assistant things like:

  • "List all inbound receipts from last month that are still unpaid."
  • "Create a draft invoice for ACME GmbH: 10 hours of consulting at €120 each."
  • "Book this bank transaction against posting account 4400."
  • "Upload this PDF receipt and assign it to the matching transaction."
  • "Show me my creditors, and add a new one for our hosting provider."

Tools are generated automatically from the official API and grouped into 🟢 read-only, 🟡 write, and 🔴 destructive — so a well-behaved host can treat each group differently.

How it works

Claude / Cursor / any MCP client  ──MCP──►  this server  ──HTTPS──►  BuchhaltungsButler API (cloud)

The server parses the bundled OpenAPI spec into MCP tools (resolving $ref batch payloads and stripping HTML from descriptions), tags each with its safety category, and injects your Basic-auth credentials and api_key on every outgoing request. Your credentials stay in the server's environment — the model never sees or handles them.

Requirements

  • A BuchhaltungsButler account with API access — an API Client + API Secret (Settings → API) and a customer api_key (see Get your API credentials).
  • Docker (Docker Desktop on macOS/Windows) for the quick start below — or Node.js ≥ 18 to run from source.

Quick start (Docker)

1. Add your credentials. Copy the example config and fill it in:

cp .env.example .env
# edit .env → set BB_API_CLIENT, BB_API_SECRET, BB_API_KEY
#           → set MCP_AUTH_TOKEN to a long random string if reachable beyond localhost

2. Start the server:

docker compose up -d --build

3. Confirm it's running:

curl -s http://localhost:3000/health     # → {"status":"ok","server":"buchhaltungsbutler-mcp"}

4. Connect your MCP client. Remote endpoints are added in Claude as a custom connector (Settings → Connectors), or bridged locally with mcp-remote. Add this under mcpServers in your client config, then fully quit and reopen the app:

{
  "mcpServers": {
    "buchhaltungsbutler": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "http://localhost:3000/mcp",
        "--header", "Authorization: Bearer YOUR_MCP_AUTH_TOKEN"
      ]
    }
  }
}

(Drop the --header line if you left MCP_AUTH_TOKEN empty.)

Prefer a prebuilt image?

Every push to main publishes a ready-to-run image to the GitHub Container Registry, so you can skip the local build entirely:

docker run -d --name buchhaltungsbutler-mcp -p 3000:3000 --env-file .env \
  ghcr.io/ohneben/buchhaltungsbutler-mcp:latest

Get your API credentials

BuchhaltungsButler uses two layers of authentication (see the official docs):

  1. HTTP Basic auth — an API Client + API Secret, your global API credentials. Find or create them in BuchhaltungsButler under Settings → API.
  2. api_key — identifies which customer account a request acts on. It lives in that customer's company-data settings.

Put all three in .env. The server injects them on every request, so your assistant never sees them. A single tool call may optionally pass its own api_key to target a different customer.

Configuration

Everything is set in .env (copied from .env.example):

Variable Required Default Description
BB_API_CLIENT ✅ — API Client (Basic-auth username)
BB_API_SECRET ✅ — API Secret (Basic-auth password)
BB_API_KEY ✅ — Default customer api_key
MCP_TRANSPORT — stdio stdio or http (the Docker image defaults to http)
PORT — 3000 HTTP listen port
HOST — 0.0.0.0 HTTP bind address
MCP_HTTP_PATH — /mcp HTTP MCP route
MCP_AUTH_TOKEN — (off) Require Authorization: Bearer <token> on /mcp
BB_RATE_LIMIT — 90 Client-side requests/minute cap
BB_BASE_URL — (from spec) Override the API base URL

After changing .env, reload with docker compose up -d --force-recreate.

Tool safety categories

Each tool's description starts with one of these banners and carries the matching MCP annotations:

Banner Count readOnlyHint destructiveHint Meaning
🟢 READ-ONLY 12 true false Fetches data only. Safe.
🟡 WRITE · creates data 22 false false Creates records (not idempotent — may duplicate).
🟡 WRITE · updates data 4 false false Modifies existing master data in place.
🟡 WRITE · links/unlinks 4 false false Assigns/unassigns receipt ↔ transaction. Reversible.
🟡 WRITE · reverts state 4 false false Un-confirms postings / restores receipts. Reversible.
🔴 DESTRUCTIVE · deletes 2 false true Deletes a record. Confirm first.

Hosts that respect annotations (Claude included) can require confirmation for destructiveHint tools and trust readOnlyHint tools automatically.

Run npm run list-tools (no credentials needed) to print the full catalog at any time.

<details> <summary><strong>🟢 READ-ONLY (12)</strong></summary>

Tool Endpoint
accounts_get POST /accounts/get
cost_locations_get POST /cost-locations/get
postings_get POST /postings/get
receipts_get POST /receipts/get
receipts_get_id_by_customer POST /receipts/get/id_by_customer
receipts_assigned_transactions_get POST /receipts/assigned-transactions/get
transactions_get POST /transactions/get
transactions_get_id_by_customer POST /transactions/get/id_by_customer
transactions_assigned_receipts_get POST /transactions/assigned-receipts/get
settings_get_creditors POST /settings/get/creditors
settings_get_debtors POST /settings/get/debtors
settings_get_postingaccounts POST /settings/get/postingaccounts
</details>

<details> <summary><strong>🟡 WRITE · creates data (22)</strong></summary>

Tool Endpoint
accounts_add POST /accounts/add
comments_add POST /comments/add
cost_locations_add POST /cost-locations/add
invoices_create POST /invoices/create
invoices_create_draft POST /invoices/create/draft
invoices_create_e_invoice POST /invoices/create/e-invoice
postings_add_free POST /postings/add/free
postings_add_receipt POST /postings/add/receipt
postings_add_transaction POST /postings/add/transaction
postings_add_batch_free POST /postings/add-batch/free
postings_add_batch_receipts POST /postings/add-batch/receipts
postings_add_batch_transactions POST /postings/add-batch/transactions
receipts_add POST /receipts/add
receipts_addBatch POST /receipts/addBatch
receipts_upload POST /receipts/upload
settings_add_creditor POST /settings/add/creditor
settings_add_debtor POST /settings/add/debtor
settings_add_postingaccount POST /settings/add/postingaccount
settings_add_batch_creditors POST /settings/add-batch/creditors
settings_add_batch_debtors POST /settings/add-batch/debtors
transactions_add POST /transactions/add
transactions_addBatch POST /transactions/addBatch
</details>

<details> <summary><strong>🟡 WRITE · updates (4) · links (4) · reverts (4)</strong></summary>

Tool Endpoint Sub-category
cost_locations_update POST /cost-locations/update update
settings_update_creditor POST /settings/update/creditor update
settings_update_debtor POST /settings/update/debtor update
settings_update_postingaccount POST /settings/update/postingaccount update
transactions_assign_receipt POST /transactions/assign/receipt link
transactions_assign_batch_receipt POST /transactions/assign-batch/receipt link
transactions_unassign_receipt POST /transactions/unassign/receipt link
postings_assign_receipt_to_free_posting POST /postings/assign/receipt-to-free-posting link
postings_unconfirm_free POST /postings/unconfirm/free revert
postings_unconfirm_receipt POST /postings/unconfirm/receipt revert
postings_unconfirm_transaction POST /postings/unconfirm/transaction revert
receipts_restore_id_by_customer POST /receipts/restore/id_by_customer revert
</details>

<details> <summary><strong>🔴 DESTRUCTIVE · deletes (2)</strong></summary>

Tool Endpoint Note
receipts_delete_id_by_customer POST /receipts/delete/id_by_customer Restorable via receipts_restore_id_by_customer
cost_locations_delete POST /cost-locations/delete Not restorable
</details>

Run from source (stdio, no Docker)

Prefer the classic stdio mode for Claude Desktop? Build it locally:

npm install
npm run build

Then point Claude Desktop at the compiled entrypoint in claude_desktop_config.json:

{
  "mcpServers": {
    "buchhaltungsbutler": {
      "command": "node",
      "args": ["/ABSOLUTE/PATH/Buchhaltungsbutler MCP/dist/index.js"],
      "env": {
        "MCP_TRANSPORT": "stdio",
        "BB_API_CLIENT": "your-api-client",
        "BB_API_SECRET": "your-api-secret",
        "BB_API_KEY": "your-customer-api-key"
      }
    }
  }
}

Or run the container over stdio instead:

{
  "mcpServers": {
    "buchhaltungsbutler": {
      "command": "docker",
      "args": [
        "run", "-i", "--rm",
        "-e", "MCP_TRANSPORT=stdio",
        "-e", "BB_API_CLIENT", "-e", "BB_API_SECRET", "-e", "BB_API_KEY",
        "buchhaltungsbutler-mcp:latest"
      ],
      "env": {
        "BB_API_CLIENT": "your-api-client",
        "BB_API_SECRET": "your-api-secret",
        "BB_API_KEY": "your-customer-api-key"
      }
    }
  }
}

(Build the image first: docker build -t buchhaltungsbutler-mcp:latest .)

Keeping the spec current

The bundled spec.json is the official BuchhaltungsButler v1 OpenAPI spec — the source of truth for the tools. To refresh against a newer API version:

curl -s https://app.buchhaltungsbutler.de/docs/api/v1.de.json -o spec.json
npm run build

New paths are picked up automatically; add them to PATH_CATEGORY in src/categories.ts so they get the correct safety category (unmapped paths fall back to the safe-but-conservative create category).

Development

npm install
npm run build      # compile TypeScript → dist/
npm test           # run the Vitest suite
npm run list-tools # print the categorized tool catalog (no credentials needed)

CI builds and tests every push across Node 20 and 22; pushes to main also publish a Docker image to the GitHub Container Registry.

Notes & conventions

  • Dates: YYYY-MM-DD. Amounts: dot decimal separator (e.g. -12.30).
  • File uploads (receipts_upload, receipts_add, receipts_addBatch): pass the file as a base64 string in the file field.
  • Paging: most get tools accept limit and offset.
  • Batch tools take arrays of objects; item schemas are resolved from the spec definitions and shown to the model.
  • Rate limit: BuchhaltungsButler allows 100 requests/customer/minute; the server self-throttles at BB_RATE_LIMIT (default 90) to stay safely under it.

Security

  • Your API credentials live only in .env, which is git-ignored. Never commit real secrets. If any leak, rotate them in BuchhaltungsButler → Settings → API.
  • The HTTP endpoint is unauthenticated by default (fine on localhost). To expose it beyond your machine, set MCP_AUTH_TOKEN and send it as an Authorization: Bearer <token> header — ideally behind TLS.

See SECURITY.md for the full policy and how to report a vulnerability.

Credits & license

An unofficial community integration for BuchhaltungsButler; not affiliated with or endorsed by BuchhaltungsButler. Built on the Model Context Protocol. Licensed under the MIT License.

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选