Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/michielinksee/bantouWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/plugins/michielinksee/bantou/claude-for-jp-accounting)<a href="https://agentmods.dev/plugins/michielinksee/bantou/claude-for-jp-accounting"><img src="https://agentmods.dev/badge/plugins/michielinksee/bantou/claude-for-jp-accounting/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/plugins/michielinksee/bantou/claude-for-jp-accounting"><img src="https://agentmods.dev/badge/plugins/michielinksee/bantou/claude-for-jp-accounting.svg" alt="Reviewed on agentmods" width="80" height="20"></a>Grade A, and why
bantou scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
{
"name": "bantou",
"displayName": "番頭 — AI仕訳アシスタント",
"version": "1.0.0",
"description": "AI-powered bookkeeping automation for Japanese tax accountants. Supports freee, Yayoi, MoneyForward, TKC.",
"author": {
"name": "Synapse Arrows PTE. LTD.",
"url": "https://github.com/michielinksee"
},
"homepage": "https://github.com/michielinksee/bantou",
"license": "Apache-2.0",
"mcpServers": {
"bantou": {
"command": "npx",
"args": ["-y", "@kansei-link/bantou@latest"]
}
},
"userConfig": {
"firm_name": {
"title": "事務所名",
"type": "string",
"description": "事務所名（例：山田税理士事務所）",
"required": true
},
"primary_software": {
"title": "メイン会計ソフト",
"type": "string",
"description": "メインの会計ソフト（freee / yayoi / mf / tkc / mixed）",
"default": "mixed"
},
"client_count": {
"title": "顧問先数",
"type": "number",
"description": "顧問先数の概算",
"default": 30
}
}
}
What it installs
The manifest is a name and a version. 5 skills, 6 commands, 1 MCP server travel with it, and installing the plugin installs all of them — 235 tokens a session between them. Each is measured on its own page, and each can be installed alone.
- Skill correction-memory A 29 tokens
- Skill nightly-batch A 31 tokens
- Skill tax-classifier A 33 tokens
- Skill csv-importer A 31 tokens
- Skill monthly-closer A 25 tokens
- Command correct A 13 tokens
- Command nightly A 15 tokens
- Command setup A 16 tokens
- Command classify A 16 tokens
- Command import A 14 tokens
- Command report A 12 tokens
- MCP server bantou A not measured
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 39 lines scan A 9490a60394b3
bantou is a plugin published in the GitHub repository michielinksee/bantou (2 stars, last pushed 21d ago), licensed MIT. Its token cost is not measured: this kind of file is read by the harness, not the model. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other plugins, from other repositories
countbean
Hosted plain-text accounting for Claude Code. Initialise a git-backed Beancount cloud book, let Claude ingest bank statements and receipts into validated double-entry transactions, and generate HTML/Excel reports.
jaz-ai
Complete AI agent stack for Jaz Accounting. 369 tools, 159 API rules, 13 IFRS recipes, 13 calculators, 12 job playbooks. Jaz Kit multi-organization workspaces with resumable closes. Data migration (Xero/QB/Sage/MYOB). Singapore Form C-S tax. For Claude Code, Copilot, Cursor, Codex, and any MCP-compatible agent.
jaz-ai
Accounting tools for AI agents and accountants: 159 API rules, 13 calculators, 13 IFRS recipes, 12 jobs.
ledger
Execution as Proof for money — the deterministic correctness layer AI agents call. Exact Money.from (no floats), kernel-enforced double-entry via JournalEntry + validateEntry + Ledger.apply, audit-hashed ledgers, canon citations. Provides /ledger- skills and slash commands.
mainbook-bank-statement-converter
Convert customer-supplied PDF bank statements to checked JSON, Excel, or CSV with balance validation through MainBook's official MCP server. Sign in with your MainBook account; no API key is copied into the configuration.
accounting-engagement
One-install umbrella for throw-work accounting agents: full engagement pipeline, all stage skills synced from modular plugins, SessionStart resume hook, and artifact schemas.