fintable_list_accountsLABSCO SUMMARY
The bulk setup Fintable's wizard makes you do one at a time.
Setting up a chart of categories and the rules that fill them is inherently a list operation, and doing it through a three-page wizard twenty times is the reason people give up on categorization. The trade is that it rides your browser session against an unofficial surface, so a change on Fintable's side can break it without warning.
What it is
A community-built connection to fintable.io, driving the same account you use in the browser. Categories, rules, transactions and connected bank accounts are all reachable, including the bulk operations the web wizard does not offer.
What you get
- Connected bank accounts with their balances, and a sync triggered through Plaid
- Transactions listed and searched, with filtering
- Categories created one at a time or up to fifty at once
- Categorization rules created singly or in bulk, listed with pagination, and deleted
- All rules run against uncategorized transactions in one call
- Updates pushed back out to Airtable and Google Sheets
Requirements
A fintable.io account with bank accounts connected, and your browser session — this is an unofficial project that talks to Fintable's own web backend rather than a public API. Python 3.10 or newer.
Setup effort
One command — uvx fintable-mcp
USE CASES
01
Setting up a whole chart of categories at once
Describe the categories you want in one sentence and have them created in a single call rather than a wizard run per category.
02
Teaching it how to sort your spending
Create the rules that map a merchant to a category — several at a time — then read the rule list back to check nothing overlaps.
03
Getting last month's transactions categorized
Run every rule against the uncategorized transactions in one pass, then list what is left to see which cases still need a rule.
04
Answering a question about an account balance
Read the connected accounts with their balances, and trigger a sync first if the figures look older than the question deserves.
05
Cleaning up rules that misfire
Page through the rule list, find the one that is catching too much, delete it and put a narrower one in its place.
06
Getting the corrected data back into your spreadsheets
Push the updates out to Airtable and Google Sheets so the reports built on top of them reflect the recategorization.
12 tools
Listed in project docs. Your client may show a different list.
fintable_list_categoriesfintable_list_rulesfintable_list_transactionsfintable_create_categoryfintable_create_bulk_categoriesfintable_create_rulefintable_create_bulk_rulesfintable_run_all_rulesfintable_delete_rulefintable_sync_accountsfintable_resync_spreadsheetsBEFORE YOU INSTALL
What to know first
Tools you get
12 tools named in the project's documentation.
Requires
- uvx on PATH
- Python 3.10+
Optional
- FINTABLE_SESSION_COOKIE in your environment
- FINTABLE_COOKIES in your environment
INSTALLATION
Connect your client
Server setup
TERMINAL
uvx fintable-mcpMCP Clients
Claude Desktop
CLAUDE DESKTOP · TERMINAL
git clone https://github.com/jasoncbraatz/fintable-mcp.git
cd fintable-mcp
pip install -r requirements.txtCLAUDE DESKTOP · JSON
{
"mcpServers": {
"fintable": {
"command": "python",
"args": [
"<YOUR_PATH_TO_FINTABLE_MCP>/fintable_mcp.py"
],
"env": {
"FINTABLE_COOKIES": "<YOUR_FINTABLE_COOKIES>"
}
}
}
}This card runs the script from a clone of the repository: <YOUR_PATH_TO_FINTABLE_MCP> is the full path of the fintable-mcp folder you cloned. The uvx fintable-mcp route above runs the published PyPI package instead.
The configuration on this page does not set FINTABLE_SESSION_COOKIE; the Claude Desktop card sets FINTABLE_COOKIES instead. The README calls the session cookie the simplest manual option, as an environment variable in the same env block. With rookiepy installed you need neither, because cookies are read from Chrome on every start.
