Bank CSV Analyst: Spending, Cashflow & Subscriptions

    1

    Turns your own bank or card CSV exports into spending by category, monthly cashflow, savings rate and recurring charges, computed by a local script.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Bank CSV Analyst: Spending, Cashflow & Subscriptions

    Example session with this skill installed

    Here are my checking and credit card exports for the last three months. Where is my money going, and what am I paying for every month?

    • Read your context and instructions
    • Compiled the bank csv analyst
    • Generated the data export

    **Plan:** read both exports, save your merchant map as `categories.json` next to them, and report income, spending by category, monthly cashflow and recurring charges. Card-payment transfers are kept out of income and spending. ### 1. Reading the files (`analyze.py --inspect checking.csv card.csv`) - `checking.csv`: header found below the bank's preamble lines, with Debit and Credit columns and US dates (`%m/%d/%Y`, not ambiguous). 25 rows. - `card.csv`: one Amount column, but 97.7% of amounts are positive, so purchases are shown as positive. It runs with `--flip-sign card.csv`. ### 2. First pass: 5 merchants uncategorized ($1,851.07) The starter map didn't know FRESHMART GROCERY, BEAN THERE CAFE, NOODLE HOUSE, STREAMBOX and CLOUDNOTES PRO, so I added them to your `categories.json` and ran again. Nothing is uncategorized now. ### 3. Summary, 1 Jan to 30 Mar 2026 | | Amount | |---|---| | Income | $18,720.00 | | Spending | $8,315.25 | | Net | $10,404.75 | | Savings rate | 55.6% | 7 transfers (savings, card payments) were excluded from both sides. | Category | Spent | Share | |---|---|---| | Rent | $5,550.00 | 66.7% | | Groceries | $1,581.86 | 19.0% | | Utilities | $480.38 | 5.8% | | Dining | $273.21 | 3.3% | | Subscriptions | $212.94 | 2.6% | | Cash | $120.00 | 1.4% | | Transport | $60.86 | 0.7% | | Fees | $36.00 | 0.4% | **Monthly net was steady:** $3,473.66, $3,427.71 and $3,503.38. ### 4. Charges that repeat every month | Merchant | Typical charge | |---|---| | MAPLE COURT APTS RENT | $1,850.00 | | CITY ELECTRIC CO | $90.31 | | BRIGHTLINE INTERNET | $65.00 | | PEAK FITNESS GYM | $45.00 | | STREAMBOX MONTHLY | $15.99 | | MONTHLY SERVICE FEE | $12.00 | | CLOUDNOTES PRO SUBSCRIPTION | $9.99 | The gym, streaming and note-app subscriptions come to $70.98 a month. The $12 monthly bank fee is easy to miss. *Auditing recurring subscriptions* in the reference covers how to review charges like these. ### Next steps - Check whether you still use each subscription, and whether the bank fee can be waived. - Rent is two thirds of spending. The *50/30/20 guideline* in the reference is a general way to judge that. It is not personal advice. - Want a goal, such as an emergency fund by December? Give me the target and date, and I'll check it against your $3,468 average monthly net.

    checking.csv

    CSV · data export

    Generated

    Example file from a real run - the skill writes it into your workspace.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Drop in your bank or credit-card CSV exports, however messy, and get an honest picture of where your money goes. Every number comes from a local script, not from the model's guess. ## What it does - **Reads messy exports:** bank preamble lines, US or day-first dates (it flags when the order is ambiguous), separate debit and credit columns, currency symbols, thousands separators, decimal commas, and card exports that show purchases as positive. - **Categorizes transactions** with a starter merchant map of about 60 entries that you extend. Anything it can't place is listed, so the agent adds it to your map and runs again. - **Reports** income, spending, net and savings rate; spending by category with shares; monthly cashflow; and charges that repeat every month, such as subscriptions, bills and bank fees. - **Keeps transfers out.** Card payments and moves to savings don't count as income or spending. - **Explains with sources.** The agent adds general budgeting guidance from a short cited reference (emergency fund, 50/30/20, savings rate, subscription audits). It never gives personal financial advice. ## Private by design Python standard library only, no network access. Account numbers are masked to their last four digits, and your files never leave your machine.

    One currency per run. No tax, investment or loan calculations. Recurring detection needs 3 or more months of a steady charge, so annual renewals and widely varying bills are missed. Duplicate rows from overlapping exports of the same account are not removed. Requires Python 3.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 9 days ago

    • Free with an Agensi account

    Requires

    Python 3

    Permissions

    Terminal / Shell
    Read Files
    Write Files

    Runs the bundled scripts/analyze.py locally on the CSV files the user names. Reads only those files, writes a categories.json or goals.json next to them only after asking, and makes no network requests.

    Listed9 days ago

    What's inside

    Frequently Asked Questions

    The skill uses a specialized Python engine to perform precise arithmetic and data normalization, preventing the common calculation errors and hallucinations that occur when an LLM tries to process long lists of transactions directly.

    You will receive the bank CSV analyst skill package, including the Python-based analytical engine, a pre-configured category mapping JSON, and a set of instructions for your AI agent to execute the workflow.

    Yes, within reason. It finds the header row below any bank preamble lines, picks the date format that parses the most values (and flags US vs day-first ambiguity), reads either one amount column or separate debit and credit columns, and handles currency symbols, thousands separators, decimal commas, parentheses and trailing minus signs. Card exports that show purchases as positive are flipped with --flip-sign.

    It lists a charge as recurring when the same merchant appears in 3 or more months and the largest charge is no more than 1.25 times the smallest, so steady bills such as an electricity bill that moves a little are caught. Annual renewals, charges seen in fewer than 3 months, and bills that swing more than that are missed.

    Any agent that can run a local Python 3 script, such as Claude Code, Codex CLI, Cursor or Gemini CLI. It reads CSV exports that have a date, a description and either an amount or debit and credit columns, which covers most bank and card downloads. Run the inspect step first to confirm the columns it picked.

    New in Finance & Accounting