Sandbox
@CPUtester5465/countbean-plugin

Claude Code plugin for Beancount bookkeeping

Countbean adds hosted plain-text accounting to Claude Code. It gives Claude a git-backed Beancount ledger, lets it ingest statements and receipts into validated double-entry entries, and produces HTML and Excel reports.

46 stars0 forksPythonUpdated 22d ago
Who it's for

Builders who use Claude Code to manage a plain-text accounting book and automate bookkeeping work.

What it delivers

You can let Claude Code turn financial documents into ledger entries and reports without leaving your book.

What it does

Beancount ledger support

Works with a git-backed Beancount ledger for plain-text double-entry bookkeeping.

Statement and receipt ingestion

Lets Claude ingest statements and receipts into validated ledger entries.

Report output

Generates HTML and Excel reports from the book.

Claude Code commands and tools

Ships plugin commands plus MCP tools that become available after install and reload.

Book connection flow

Uses `/countbean:connect` to link Claude Code to a Countbean book in the web app.

How to get it

  1. 1Run these one at a time — Claude Code reads a multi-line paste as a single slash…
    /plugin marketplace add https://github.com/CPUtester5465/countbean-plugin.git
  2. 2Run
    /plugin install countbean@countbean
  3. 3Then reload, so the running session picks up the plugin's commands and MCP tools — they…
    /reload-plugins
  4. 4Then connect it to your book. Say this to Claude with nothing after it
    /countbean:connect

README

Countbean plugin for Claude Code

Hosted plain-text accounting in Claude Code: a git-backed Beancount ledger, AI ingestion of statements and receipts into validated double-entry, and HTML/Excel reports.

Install

Run these one at a time — Claude Code reads a multi-line paste as a single slash command, so sending them together makes marketplace add swallow the install line and fail with URL rejected: Malformed input to a URL function.

/plugin marketplace add https://github.com/CPUtester5465/countbean-plugin.git
/plugin install countbean@countbean

Then reload, so the running session picks up the plugin's commands and MCP tools — they are bound when Claude Code starts, so a fresh install is not live until you do:

/reload-plugins

Use the full HTTPS URL, not the CPUtester5465/countbean-plugin shorthand — the shorthand resolves to git@github.com, which needs an SSH key and GitHub's host key already in your known_hosts.

Then connect it to your book. Say this to Claude with nothing after it:

/countbean:connect

It prints a short code and a link. Open the link, sign in at app.countbean.com, pick the book, and approve — Claude picks it up within a few seconds. There is no key to copy between windows. (If you would rather paste a key you already have, put it on the same line: /countbean:connect cbk_… bok_….)

Full documentation is in plugin/README.md.


This repository is published, not authored

The source of truth is the plugin/ directory of the Countbean monorepo. This repo is a mirror, pushed by CI on every change.

Pull requests here cannot be merged — they would be overwritten by the next publish. The mirror exists so that installing the plugin does not require access to the private monorepo, and so plugin/mcp/countbean_mcp/ledger.py can keep being byte-compared against its ledger_core original by tests that need both in one tree.

Files in the repo

Repository payload6 top-level entries
  • .claude-plugin
  • plugin
  • Dockerfile
  • LICENSE
  • pyproject.toml
  • README.md

Discussion (0)

Ask about usage, or say what you built with it

Sign in to join the discussion.

No comments yet. Be the first to say what this is good for.

More plugins

An Obsidian plugin that embeds Claude Code/Codex as an AI collaborator in your vault

15k
nagisanzenin/
engram

Evidence-based learning engine — first-principles curricula, free-recall verification with receipts, FSRS-scheduled memory, and explorable artifacts. Learn anything; keep it.

1.4k
agenticnotetaking/
arscontexta

Claude Code plugin that generates individualized knowledge systems from conversation. You describe how you think and work, have a conversation and get a complete second brain as markdown files you own.

3.5k
Aisland-SJL/
dsh-usage

🌊 Persistent dock & fully-customizable balance/usage panel for DeepSeek Harness — activity heatmap, dual-channel comparison, local-only & privacy-first

92