OpenAccountants

Open-source tax guides for AI agents, reviewed by named, licensed accountants — answers your AI can cite on the record.

Stars
★ 500
Last updated
1d ago
License
AGPL-3.0
Primary language
HTML

At a glance

How it runs
MCP serverHosted serviceLibrary / SDK
Works with
Universal · cross-platformChatGPT · Claude.ai
Cost
Free tier plus a paid hosted plan
Setup effort
Low · running in minutes
You'll need
Python (pip install openaccountants-mcp)Shell / CLINetwork accessMCP Server
Typical use
A South African freelancer asks Claude what they owe and receives an ITR12 working paper, IRP6 schedule and the reviewer's byline
Not a fit if
  • Users needing formal, signed tax advice for filing decisions
  • Teams that cannot send queries through a hosted connector or external platform
Website
openaccountants.com/connect ↗ set on GitHub
Maintenance
Commits in the last 30 dayslast commit 2026-11-20latest release v2.4.0 (2026-11-18)+1.6 ★/day over 7 days
Source review
76/100 · Good

What does this agent do, and when should you use it?

OpenAccountants is an open-source tax guide library (AGPL-3.0 for code) spanning 1,881 guides across 230 jurisdictions, of which 53 are accountant-reviewed by 44 named professionals. Every guide sits in one of two greppable states: reviewed by a named licensed accountant (CPA/CA/EA) credited in the frontmatter, or a source-cited draft awaiting professional review. You can consume it three ways: add the hosted MCP connector (https://www.openaccountants.com/api/mcp) to Claude, ChatGPT, Cursor or any MCP client; self-host the MCP server via pip (openaccountants-mcp); or download jurisdiction folders from packages/ and upload them manually. Nightly automation keeps the library honest through a maths check, a source watch, a refresh engine and a nightly exam that catches AI improvisation. Guide content carries a separate OpenAccountants Guide License v1.0 with commercial licensing options.

The repo keeps guide sources in skills/ per jurisdiction; nightly jobs generate packages/ per-country bundles, the machine-readable index.json, and the llms.txt entry point. Users point Claude/ChatGPT/Cursor at the hosted MCP endpoint or run pip install openaccountants-mcp as a local MCP server; when asked a question, the AI loads relevant guides (e.g. za-income-tax, za-provisional-tax) and produces working papers such as ITR12 and IRP6 schedules with the reviewing accountant's byline (e.g. Werner Britz CA(SA)). A nightly pipeline runs a maths check (tax band gaps, totals that don't total), a source watch (alerts when official pages change), a refresh engine (re-derives stalest guides from official sources on a monthly budget, refusing to publish unlinked figures), and a nightly exam (verifies AIs answered per the guides). Platform edits are committed back under the author's own GitHub identity, and merged PRs flow back into the platform.

  1. A South African freelancer asks Claude what they owe and receives an ITR12 working paper, IRP6 schedule and the reviewer's byline
  2. A developer in Cursor queries a 2026 combined sales tax rate via the MCP endpoint instead of letting the model guess from training data
  3. An accountant authors a guide for their home jurisdiction via openaccountants.com/skills/new, publishing it credited under their name
  4. A licensed CPA/CA/EA reviews guides in their jurisdiction, putting their name, credential and review date on every AI answer that cites them
  5. A privacy-sensitive team downloads a jurisdiction folder from packages/ and uploads index.json and guide files to their own AI without the hosted connector

Key setup facts

Pulled from the install and usage notes below. The project README has the current values.

Install / run
  • pip install openaccountants-mcp
Config files
index.json

How do you install or deploy this agent?

Fastest path: add the hosted connector in your MCP client:

text

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

Guided setup: openaccountants.com/connect. Self-hosted path:

bash

pip install openaccountants-mcp

Manual path: download your jurisdiction's folder from the repo's packages/ directory and upload the files (including index.json) to your AI.

How do you use this agent?

Once connected, just ask, e.g.:

text
What's the combined sales tax rate in Manatee County, Florida for 2026?

The AI loads the matching guides and answers citing the current Guide and the reviewing accountant. Have a qualified professional review outputs before filing, payment, or action.

What are this agent's strengths and limitations?

Pros
  • Differentiator: each reviewed guide carries a named accountant's name, credential and review date, publicly on the record in VERIFIERS.md
  • Broad coverage with flexible delivery: 1,881 guides across 230 jurisdictions, consumable via hosted MCP endpoint, pip self-hosted server, or manual files
  • Actively maintained accuracy: nightly maths check, source watch, refresh engine and a nightly exam that catches and fixes AI improvisation at the source
Limitations
  • Only 53 of 1,881 guides are accountant-reviewed; most content is still source-cited draft without professional sign-off
  • Explicitly general reference, not advice — outputs must be reviewed by a qualified professional before any filing or payment
  • Guide content uses the OpenAccountants Guide License v1.0 (distinct from the AGPL-3.0 code license), with commercial use requiring separate licensing
  • The hosted connector sends queries to openaccountants.com; teams that cannot route data through the vendor's cloud must self-host or use manual uploads

How does this agent compare with similar options?

Key facts side by side with the most closely related agents.

Agent Source review Form / cost Stars Updated Language Full support on
OpenAccountants This agent 76 · Good MCP serverFreemium ★ 500 1d ago HTML ChatGPT · Claude.ai
MCP Brasil 55 · Major gaps MCP serverFree ★ 1.9k 3mo ago Python Claude Code · Claude.ai
webclaw Web Extraction Engine 80 · Good CLIFreemium ★ 2.4k 2d ago Rust Codex · Claude Code · OpenAI API · Claude API
TradeMemory Protocol 74 · Some gaps MCP serverFree ★ 1.4k 1mo ago Python Claude Code · Claude.ai

How does FollowAgents rate this agent?

FollowAgents source review · FARS-2.1
Good
76/ 100 5-point scale 3.8 / 5
This review covers the source as of Nov 08, 2026. The project has shipped 2 stable release(s) since, which have not been re-reviewed. The score and verdict describe the reviewed version only. Recent releases ↓
Trust 22/29
Reliability 8/14
Adaptability 16/18
Convention 14/18
Effectiveness 12/13
Verifiability 4/8
Why each dimension lost points
Trust22 / 29 · 3.8/5

The MCP code is read-only, constrains paths to prevent escapes, defaults HTTP binding to 127.0.0.1, and states that hosted mode calls website APIs; this supports least privilege, data-flow transparency, and low external impact. Tax skills ask for client facts and list refusal cases, but the excerpts do not show a consistent pre-action confirmation mechanism. The security policy gives a reporting channel, while dependency pinning or vulnerability mitigation evidence is not shown.

Reliability8 / 14 · 2.9/5

The server has explicit handling for path escapes, duplicate slugs, and malformed frontmatter, and test files document freshness and duplicate-handling regression cases. However, README coverage figures (2,132/249) differ from the supplied description (2,250/266) and its 190+ claim, with potentially different counting scopes unexplained; the skill samples are also truncated. This lowers self-consistency. Input errors have a user-facing message design, while dependency availability is only partly documented.

Adaptability16 / 18 · 4.4/5

The materials cover many jurisdictions and tax scenarios, state audiences, inputs, refusal cases, and related-skill dependencies. MCP supports bundled packages, stdio, configurable HTTP, and several client setup paths. Trigger descriptions include concrete keywords, though some sample language such as “any question” and “ALWAYS read” is broad.

Convention14 / 18 · 3.9/5

The repository explains the separation of source skills, generated packages, inventory, server, and documentation, and gives installation, self-hosting, and manual-file paths. The software/content license split is detailed; examples, quality tiers, limitations, and security-reporting routes are clear. Nightly sync, monthly refresh, and annual-update processes are described, but the supplied sample does not include a full changelog or identify one accountable maintainer.

Effectiveness12 / 13 · 4.6/5

Skills offer structured steps, rate tables, client questions, refusal catalogues, and source notes; the calculator sample documents inputs and result fields, and MCP supports discovery and retrieval. The likely value is substantial, but broad scope and explicit outdated-or-wrong-data cautions mean outputs need professional review. Free connection and self-hosting are available, with commercial licensing also described.

Verifiability4 / 8 · 2.5/5

Sample skills connect tax claims to statutory provisions, and the Andorra example says it uses consolidated texts while identifying the official gazette text as controlling. The repository also describes independent checks and source monitoring. Still, many specific figures in the excerpts lack individually inspectable citations, and the supplied evidence offers limited direct corroboration from independent sources. Claims and scope are generally distinguished, but inference is not consistently labeled.

Risks and how to mitigate them
  • Tax figures may be incomplete, outdated, or wrong; have a qualified professional check official sources and the specific facts before filing, paying, or acting.
  • README coverage figures differ from the task description, and excerpts are truncated; do not treat them as confirmation of total coverage or current status for every skill.
  • Code and tax skills have different licenses; check the applicable path-specific terms before reuse or redistribution.
Evidence confidence: Low Reviewed Nov 08, 2026 Reviewed revision 2ccff471255d New commits since this review; the score may not cover them
See the full review method →
Maintain this project? Add the FARS badge to your README

The badge shows the current review score and verdict and links back to this page. It refreshes about once an hour after a re-review.

FARS

[![FARS review](https://followagents.com/badge/openaccountants.svg)](https://followagents.com/en/agents/openaccountants)
<a href="https://followagents.com/en/agents/openaccountants"><img src="https://followagents.com/badge/openaccountants.svg" alt="FARS review"></a>

Recent releases

Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.

  1. v2.4.0 v2.4.0: 283 places, 34 named accountants, every skill to the 2026 standard
    Release notes excerpt
    What's new since v2.3.0 (28 October) - 2,555 tax skills for 283 countries, states and territories, every figure linked to its official source. Every place in the world we cover now has at least one skill; sanctioned countries are listed as not covered. - The 2026 standard everywhere: 2,282 of the skills that no accountant has approved yet pass every check (sourced rate rows, recomputed worked examples, current tax year); 34 are still being finished. …
    View on GitHub ↗
  2. status Repository status (machine-readable)
    Release notes excerpt
    status.json is rebuilt daily and on every skills change. Not a software release.
    View on GitHub ↗
  3. v2.3.0 v2.3.0: 266 jurisdictions, 32 complete packs, tax year 2026
    Release notes excerpt
    What's new since v2.2.0 (July) - 2,250 tax skills across 266 jurisdictions, every figure cited to its official source. - 32 complete country packs (all seven tax domains: individual and corporate income tax, VAT/GST, social contributions, payroll, transfer pricing, cross-border), including most of Europe, rebuilt for tax year 2026. …
    View on GitHub ↗

FAQ

Does it cost anything?
The repo and pip package are open source (AGPL-3.0 for code); the source does not state a fee for the hosted connector. Guide content has commercial licensing options whose pricing is not stated.
Are all guides accountant-reviewed?
No. As of 2026-10-04, 53 of 1,881 guides are reviewed by 44 named accountants; the rest are source-cited drafts. Status is verifiable via frontmatter and VERIFIERS.md.
Which AI clients are supported?
The README names Claude, ChatGPT, Cursor, Windsurf or any MCP client, via the hosted endpoint or the self-hosted openaccountants-mcp server.
Can I use it fully offline?
Partially: download jurisdiction folders from packages/ and upload them to your AI for offline use of static guides, but the nightly maths checks, source watch and sync run only on the platform side.
Can I contribute?
Yes. Edit only skills/** (packages/, index.json and llms-full.txt are generated nightly and must not be edited in PRs); accountants can author or review guides on the platform, with commits attributed under their own GitHub username.
Is OpenAccountants open source?
Yes. The source code is on GitHub (openaccountants/openaccountants) under the AGPL-3.0 licence. AGPL-3.0 is a copyleft licence that also covers network use: if you run a modified version as a service, you must offer its source to users.
Is OpenAccountants safe to use?
FollowAgents' FARS review (Nov 08, 2026) scored OpenAccountants 22/29 on trust and safety. None of the eight safety controls the review checks for was missing outright. The repository has changed since that review.
Is OpenAccountants still maintained?
Yes. The latest commit on GitHub is from 2026-11-20, within the last 30 days. The latest release is v2.4.0 (2026-11-18).
Who makes OpenAccountants?
OpenAccountants is published on GitHub by the account openaccountants. FollowAgents has not verified which person or company is behind that account.
What is OpenAccountants's official website?
The project's GitHub repository lists https://www.openaccountants.com/connect as its website. That link is set by the repository owner; the source code itself is on GitHub (openaccountants/openaccountants).
Was this page helpful?
View on GitHub ↗ Install ↓

Compare agents like this one

The same FARS review applied across the shortlist this agent qualifies for.

Related concepts