openaccountants is a free, open source finance & accounting project written in HTML and released under AGPL-3.0. It has 500 GitHub stars, 74 forks and 25 open issues, and was last pushed 12 hours ago. On this registry it ranks #36 of 40 tracked projects in Finance & Accounting, with 5 head-to-head comparisons available.

What is openaccountants?

OpenAccountants is an open-source library of tax skills for AI assistants — more than 2,000 skills covering 190+ countries and states, each figure linked to the law it came from — built for accountants, tax practitioners and developers who need an AI to answer from sources rather than from memory.

What it is

OpenAccountants is a repository of agent skills: tax knowledge written from official laws, gazettes and tax authority pages, and packaged for AI assistants to load. It lives in the agent-skills and MCP ecosystem, and its skills are consumed in ChatGPT, Claude, Cursor and any MCP client. The content is licensed AGPL-3.0 and served largely as HTML, and the project is listed in the MCP Registry as io.github.openaccountants/openaccountants. The repository runs behind what the OpenAccountants plugin serves.

The problem it solves is the confident, unverifiable tax answer. A general-purpose model answers tax questions from memory, so a rate or threshold can be stated wrongly and cannot be checked; loading these skills moves the source into the open, because every figure sits in a row that links the official page it came from. The README's example is the Malta VAT skill, where economic activities consisting principally in the supply of goods carry a €35,000 registration threshold and a €28,000 exit threshold, each tied to legislation.mt. For income tax, GST and compliance, this replaces the model's recollection and the blog or firm summary that would otherwise stand in for the law.

Key capabilities

  • More than 2,000 skills across 190+ countries and states; the README counts skills for 283 countries, states and territories, with income tax, GST and compliance among the project's topics.
  • Rates tables in which every figure carries a source link, such as the Malta VAT row pairing a €35,000 registration threshold with the legislation.mt page it came from.
  • An automated figure check (docs/AUTOMATED-REVIEW.md, run by the validate.yml workflow) that looks for each rates-table figure on the page its row links to and blocks the merge when the figure is not found.
  • An evidence archive in openaccountants/evidence holding snapshots of source pages and the working ledgers behind the skills, so a figure stays traceable after a government site changes.
  • Named approval, where "Approved by [name]" appears only after a licensed accountant has read that exact text, with 34 named accountants approving in 28 countries and states and more than 100 skills approved by a named, licensed accountant.
  • Approval that clears when a skill's text changes, delivery as an MCP plugin for ChatGPT, Claude, Cursor and any MCP client, and OpenSSF Scorecard and Best Practices badges behind the repository.

Who uses it and how

  • Licensed accountants approve the skills for their own country through the accountant portal, in one click from their own GitHub account, after which their name appears beside every AI answer that uses those skills in ChatGPT, Claude and Cursor.
  • Practitioners working in ChatGPT, Claude, Cursor or another MCP client add the plugin so that rates, thresholds and deadlines arrive with citations and an approval indicator attached.
  • Contributors file corrections and new skills as pull requests, where the figure check must pass and a text change costs the skill its approval until an accountant re-reads it.
  • Reviewers use the evidence archive to trace a figure after the underlying government page has changed.

Getting started

Add the plugin from https://www.openaccountants.com/connect, or find it in the MCP Registry as io.github.openaccountants/openaccountants; the repository runs behind what the plugin serves, so the plugin is the route to the current, accountant-approved skills.

How it compares

The facts provided name no comparable or paid product that OpenAccountants replaces, and ChatGPT, Claude and Cursor appear only as clients that consume its skills. On the evidence available, it stands alone in this registry.

When to use it — and when not

It suits work where a tax figure has to be traceable and attributable to a named approver, and it does not suit anyone who wants an AI to settle a tax position without a licensed professional: OpenAccountants does not sell accountancy services, and each approving accountant works under their own licence. Two limits are visible in the facts: approvals appear only through the plugin and not in the repository, so a reader of the code alone cannot see which skills are approved, and coverage is uneven, with 34 named accountants across 28 countries and states against skills for 283. The README excerpt also describes the hosted plugin rather than any self-hosting stack, and 25 issues remain open.

project readme (upstream, from github) — read inline

OpenAccountants

Every skill here is written from the official law and the tax authority's own pages, with each figure linked to its source, and checked before it is published. Accountants write and approve these skills, but approvals, and the accountant's name, appear only through the OpenAccountants plugin (MCP). More than 100 skills are approved by a named, licensed accountant, and this repository runs behind what the plugin serves. For the current, accountant-approved version, add the plugin to your AI: https://www.openaccountants.com/connect

Tax skills for AI, written from the law and approved by named accountants. Your AI answers tax questions from memory. Add these skills and it answers from the law: every rate, threshold and deadline links to the official page it came from, and the plugin says which accountant, if any, has approved each skill.

OpenSSF Scorecard OpenSSF Best Practices Skill checks Approved by accountants MCP Registry License: AGPL-3.0 GitHub stars

OpenAccountants MCP server on Glama

34 named accountants approving skills in 28 countries and states · skills for 283 countries, states and territories · every figure linked to its official source


Accountants

A skill is Approved by [name] only when a licensed accountant read that exact text and approved it (how approval works). If the text changes, the approval is cleared until they approve again. Who approved which skill is shown on openaccountants.com and in the plugin's answers, not in this repository.

Are you an accountant? Approve the skills for your country and your name appears next to every AI answer that uses them, in ChatGPT, Claude and Cursor. It takes one click from the accountant portal, from your own GitHub account. Found something wrong? Correct it in the same place. Each accountant works under their own licence; OpenAccountants does not sell accountancy services.


Every figure has a source

A tax answer is only as good as the number in it. So every figure in a skill sits in a row that links the official page it came from, and the page can be opened and checked by anyone. A real row, from the Malta VAT skill:

Activity Registration threshold Exit threshold Source
Economic activities consisting principally in the supply of goods €35,000 €28,000 https://legislation.mt/getpdf/67691cdeb52f1d38d8a4103f

What keeps those figures honest:

  • Official sources only. Laws, gazettes and tax authority pages; never blogs or firm summaries.
  • The figure check. On every pull request, each figure in a rates table is looked for on the page its row links to. A figure that is not there blocks the merge until it is fixed or explained.
  • The evidence archive. Snapshots of the source pages and the working ledgers behind the skills live in openaccountants/evidence, so a figure can still be traced after a government site changes.
  • Worked examples that add up. Every skill's examples are recomputed on every pull request (scripts/check-examples.py).

How skills are built, checked and corrected: docs/ACCURACY-METHODOLOGY.md.


Add to your AI

Add this address to Claude, ChatGPT, Cursor, Windsurf or any AI that takes a connector:

https://www.openaccountants.com/api/mcp

Guided setup: openaccountants.com/connect. Then ask a question your AI would otherwise guess at:

"I'm a freelancer in Malta billing €40,000 this year. Do I have to register for VAT?"

Prefer self-hosting or manual files?

Also listed on the MCP Registry and Glama.


Find your country

Browse all countries, by region, with the state of each pack. United States: federal and states.

In the plugin, each skill carries one of two labels (definitions):

Label Meaning
Approved by [name] A named, licensed accountant approved this exact text. The name is shown in the plugin and on the website.
Not yet approved Built from the law and checked before merge, but no accountant has approved it yet.

⚠️ General reference, not advice. Skills may be incomplete, outdated, or wrong for your facts. Have a qualified professional review outputs before filing, payment, or action.


Contribute

Three ways in, all in CONTRIBUTING.md:

  • I am an accountant. Approve or correct the skills for your country, under your own name.
  • I can write. Take a country and write its missing skills from the law.
  • I found a mistake. Open an issue with the official source that shows it.

Need an accountant for your own bookkeeping, VAT, payroll or return? Get in touch and we will put you in touch with one in your country.


For developers

What Where
Skill source (per jurisdiction) skills/
Per-country bundles (generated) packages/
Machine-readable inventory index.json · llms.txt
What is left, per country (rebuilt daily) status.json
Python MCP server mcp/ · PyPI
Repo layout and sync docs/REPO-LAYOUT.md · docs/WEBSITE-SYNC.md

API and platform integrations: openaccountants.com/for-developers


Related open-source projects

Independent, open-source review aids and calculation tools that complement the OpenAccountants skill library:

  • payday-super-checker (Apache-2.0): Verifies superannuation contributions against Australian Payday Super deadlines (in force since 1 July 2026) and estimates SG charge exposure on late payments, with all workings and statutory assumptions disclosed.
  • div7a-loan-review (Apache-2.0): Reviews Division 7A loan terms and minimum yearly repayments against the ATO benchmark interest rate, with fabricated inputs and explicit refusal boundaries.
  • xero-trial-balance-export (Apache-2.0): Exports a Xero trial balance to CSV only when month-movement and YTD column pairs both balance exactly.
  • ato-benchmark-compare (Apache-2.0): Evaluates business P&L statements locally against the Australian Taxation Office (ATO) small business performance benchmarks, surfacing variance flags and ratio analyses.
  • OsaBooks (AGPL-3.0): Free bookkeeping for businesses, rental properties and households in Australia, New Zealand, South Korea and the United States, with Xero and QuickBooks imports. Use it as a standard bookkeeping platform, or with several AI features: coding suggestions, year-end checks against the OpenAccountants guides, and an MCP connector so Claude, ChatGPT or another AI app can read your books. Runs on your own computer or online on your own Supabase. Home page and demo.

(External tools provide computational review assistance and do not constitute tax or legal advice.)


License

Contact: info@openaccountants.com · Security policy · Cite this repo

Frequently asked questions

Is openaccountants free to use?

openaccountants is open source under the AGPL-3.0 licence. There is no licence fee and no seat count — you can self-host it or, where the project offers one, pay a vendor for a managed version instead.

What does openaccountants do?

Open tax skills for AI: 2,000+ skills for 190+ countries and states, every figure cited to the law. Works in ChatGPT, Claude, Cursor and any MCP client.

What is openaccountants written in?

openaccountants is primarily written in HTML. Its source is publicly available at https://github.com/openaccountants/openaccountants, and it has 500 GitHub stars.