About tiedout
tiedout is an MCP server and CLI for small businesses, freelancers, and bookkeepers to reconcile bank and invoice CSV files. It uses exact calculations to group results by confidence and reason, while keeping CLI processing local.
Key features
- 01Match bank CSVs with invoices or bills
- 02Group results by certainty and explain why
- 03Detect possible duplicate payments
- 04Read varied CSV formats and column names
- 05Run locally without network access
Use cases
- 01Reconcile customer payments against invoices
- 02Match outgoing payments to supplier bills
- 03Find unmatched transactions and possible duplicates
MCP support
Provides tools to reconcile two CSV files and inspect detected columns and row counts. Reconciliation results sent through MCP include names and amounts from the files and are returned to the MCP client.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
Tools
2Saved tool list · updated 2w ago
| Parameter | Type | Required | Description |
|---|---|---|---|
| mode | enum | No | receivable matches money in to customer invoices; payable matches money out to billsreceivable | payable |
| limit | integer | No | Most items per group. 0 returns only the summary |
| bank_path | string | Yes | Path to the bank statement CSV |
| date_format | enum | No | Needed only if the tool reports that dates are ambiguousdmy | mdy |
| window_days | integer | No | How long after the invoice date a payment can count, for matches without an invoice number |
| bank_columns | object | No | Column names, when detection fails, such as {"date":"Posted","amount":"Value"} |
| invoices_path | string | Yes | Path to the invoice or bill CSV |
| invoice_columns | object | No | Column names, when detection fails, such as {"number":"Doc","amount":"Gross"} |
Details
From the project READMEtiedout: bank reconciliation for CSV files
Match your bank statement to your invoices or bills, with exact math and without uploading your books anywhere. For small businesses, freelancers and bookkeepers.
Reconciling payments by hand is slow, and asking an AI model to compare long CSV files invites arithmetic mistakes. tiedout does the matching in code and sorts every line into matched, likely, review or unmatched, with the reason. A person or an agent only has to look at the exceptions. It's an MCP server and a CLI with no account, no API key and no network access.
It works with any bank or invoicing tool that exports CSV. It reads title lines, footers, separate credit and debit columns, semicolons, decimal commas, day-first dates and several languages' column names.
What you can ask your agent
- "Here's my bank export and my invoices. Who has paid?"
- "Which customers still owe me money?"
- "Match these supplier bills to my bank statement."
- "Are there any duplicate payments, or payments with no invoice?"
Install
As a Claude Code plugin, which adds the MCP server and a skill that tells Claude when to use it:
claude plugin marketplace add GillMacedo/tiedout
claude plugin install tiedout@tiedout-marketplace
For other MCP clients, run it from npm. You need Node 22.19 or newer:
{
"mcpServers": {
"tiedout": {
"command": "npx",
"args": ["-y", "tiedout", "serve"]
}
}
}
Usage
Try it on the sample files in examples/. All the names in them are made up.
$ node plugin/dist/tiedout.js match examples/bank.csv examples/invoices.csv
MATCHED (2)
2026-03-02 1200.00 TRANSFER FROM EXAMPLE SUPPLIES REF INV-1001 -> INV-1001 [invoice number and amount]
2026-03-04 450.50 PAYMENT FICTIONAL CAFE INV 1002 INV1003 -> INV-1002, INV-1003 [invoice numbers and total]
LIKELY: check these (2)
2026-03-09 975.00 SAMPLE WIDGETS ONLINE TRANSFER -> INV-1004 [amount and party name]
2026-03-16 700.00 PLACEHOLDER GARDENS -> INV-1007, INV-1008 [party name, and the amounts add up]
REVIEW: a person should decide (3)
2026-03-11 590.00 ACME TEST CO INV-1005 -> INV-1005 [names INV-1005, but short by 10.00]
2026-03-18 1200.00 TRANSFER FROM EXAMPLE SUPPLIES REF INV-1001 [names INV-1001, which is already settled: possible duplicate payment]
2026-03-12 310.00 DEPOSIT 3301 -> INV-1006 [amount only]
UNMATCHED BANK LINES (1)
2026-03-20 42.00 UNKNOWN SENDER
OPEN INVOICES (1)
INV-1009 125.00 Mock Logistics due 2026-03-28
SKIPPED bank row 12: no valid date
2 matched, 2 likely, 3 to review, 1 unmatched, 1 open invoices
Use --payable to match money out to bills. Run tiedout with no arguments to see all options, including --json and naming columns by hand.
How it decides
| Group | Rule |
|---|---|
| Matched | The bank line names the invoices, and the amounts agree exactly |
| Likely | The party name appears and the amount fits only one option, or a set of that party's invoices adds up in only one way |
| Review | A named invoice with the wrong amount, a possible duplicate, a tie, or an amount with nothing else to support it |
It never breaks a tie, and it never reports a row as gone without saying why.
MCP tools
| Tool | What it does |
|---|---|
reconcile_match | Matches two CSV files and returns the groups, with a limit for large results |
reconcile_inspect_csv | Shows detected columns and row counts, without row contents |
Both are read-only. reconcile_match returns names and amounts from your files to the MCP client, so they reach whichever AI model you use. The tiedout CLI keeps everything on your machine.
Limits
- This helps with bookkeeping and isn't accounting advice. Check results before relying on them.
- It reads CSV only, so PDF statements need converting first.
- It assumes two decimal places and doesn't convert currencies. Lines in different currencies never match.
- It treats each file as the full list of open items. It doesn't track partial payments across runs.
- So far it has been tested on made-up data in common export formats, not on every bank's real exports. If yours fails, please open an issue with the column headers, not your data.
FAQ
How do I reconcile a bank statement with invoices automatically? Export both as CSV and run npx tiedout match bank.csv invoices.csv. Matches come out with the reason, and anything uncertain goes to review.
Is my financial data uploaded? The tool has no network code. The CLI keeps everything on your machine. Through MCP, results go to whichever AI model your client uses.
Does it work with QuickBooks, Xero or my bank? It reads CSV exports rather than connecting to them. Most accounting tools and banks can export CSV. It hasn't been tested on each one's real exports yet.
Can it match accounts payable? Yes. Use --payable to match money out to bills.
Contributing
Contributions are welcome, from people and from people working with AI agents. See CONTRIBUTING.md. Good first issues:
- Add column names for more languages and accounting tools
- Add
--outto write results to a CSV for a spreadsheet - Support OFX and QIF bank files
- Track partial payments across several bank lines
- Add a plugin icon
How pull requests work
- Open a PR. Fork the repo, work on a branch, and keep it to one change. Write a title that says what changes for users, because it becomes the changelog line:
feat:goes under Added andfix:under Fixed. - Automatic check. Code changes need tests, and
make testmust pass. Changes over 30 lines need a linked open issue (Closes #<number>), and a PR can have at most 800 changed lines. If something is missing, a comment tells you what to fix. - Review. Claude reviews each PR automatically. Code PRs also get a security review and are tested in a sandbox. The review then merges the PR or asks for changes. PRs that change CI, agent settings, npm or Python dependencies, vendored code or build files can't be merged this way, and a comment says so. The same goes for anything the security review doesn't clear.
- Release. A merge that changes what users install becomes a new version within about two hours, with a GitHub Release and, if the project has one, its npm package. Docs and CI changes wait for the next release.
AI disclosure
This project was largely generated with Claude Code and reviewed by its maintainer.
About me
I'm Giulliano Macedo, and I build developer tools at GoTech Solutions. Follow along or get in touch:
- GitHub: GillMacedo
- LinkedIn: giulliano
- X: @giumacedo
If tiedout saved you an afternoon, a star helps others find it.
MIT licensed.
Built by Giulliano Macedo, GoTech Solutions · LinkedIn · X
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.