Import a Rocket Money transaction CSV export into the local store. Idempotent: re-importing the same or an updated export dedupes by a stable row hash, so only genuinely new transactions are added. Args: path (absolute path to the .csv). Returns counts (parsed/inserted/skipped), the date span, and any parsing warnings.

Summarize what's in the store: total transaction count, date span, last import time, and per-category inflow/outflow totals across all history. Call this first to know what data is available. No args.

Search transactions with filters. All optional; combine freely. Returns newest-first, paginated. Amounts are signed (positive = money in, negative = money out). Args: from/to (YYYY-MM-DD), category, merchant (substring on name/descriptor), min_amount/max_amount (by absolute value), flow ('in'|'out'|'all'), limit (1-500, default 100), offset. Returns { total, count, offset, has_more, transactions: [{id, txn_date, amount, description, category, account, institution}] }.

Total inflow and outflow grouped by category over an optional date range (defaults to all history). Sorted by outflow descending. Args: from/to (YYYY-MM-DD). Returns { range, categories: [{category, inflow, outflow, count}], total_outflow, total_inflow, net }.

Detect recurring charges (subscriptions, memberships, bills) from transaction history using merchant-key normalization and cadence analysis. Flags price hikes vs the prior baseline. Args: none (analyzes all history). Returns { totals: {monthly, annual, count, hikes}, subscriptions: [{label, category, cadence, lastAmount, monthlyCost, occurrences, lastDate, nextDate, daysUntilNext, priceHike?}] }.

Project cash flow over the next N days from detected recurring bills (outflows) and recurring income (inflows, cadence-aware). Args: starting_balance (current available cash; default 0), window_days (7-90, default 30). Returns a day-by-day projection with the running balance and a flag for the lowest point. Note: starting_balance isn't in the CSV — pass your real current balance for an accurate low-point.

Every charge from a given merchant (substring match on name/descriptor), with a spend summary. Useful for 'how much have I paid X' and spotting creep. Args: merchant (required), from/to (optional). Returns { merchant, count, total_spent, total_received, first_date, last_date, average, transactions: [...] }.

Simulate a debt-payoff plan given a list of debts and a monthly budget. Strategy 'avalanche' (highest APR first) or 'snowball' (smallest balance first), or 'compare' to run both and show interest/time saved. Debts are provided by you (not read from the CSV). Args: debts: [{id, name, balance, apr, minPayment}], monthly_budget, strategy. Returns the plan(s): months to debt-free, total interest, debt-free date, per-debt payoff month.

Connect a bank via SimpleFIN Bridge for real-time balances and ongoing transaction sync. Get a one-time setup token from bridge.simplefin.org (My Account → generate a token) and pass it here. Claims the token into a durable access URL stored locally, then runs an initial 365-day sync. Idempotent transactions (dedupe by provider id). Args: setup_token. Returns account balances + import counts.

Live per-account balances and net worth via SimpleFIN (fetched fresh on each call, then cached). Credit-card and loan balances are shown as positive amounts OWED and subtracted from net worth. If the live fetch fails, returns the last cached balances with a note. Requires finance_link_bank first. No args. Returns { accounts: [{name, institution, type, balance, currency, as_of}], assets, liabilities, net_worth, live }.

Pull new transactions and refresh balances from SimpleFIN. By default fetches from the day after the latest transaction already in the store (so it won't re-pull or double-count history from your CSV) — pass days_back to force a wider window (e.g. 365 for a full refresh). Transactions dedupe by provider id. Requires finance_link_bank first. Args: days_back (optional, 1-730). Returns { accounts, newTransactions, store_span }.

Forget the SimpleFIN connection (clears the stored access URL). Imported transactions and cached balances are kept. No args.

Create a rule that assigns a category to any transaction whose merchant name or bank descriptor contains a given substring — and immediately recategorize all matching transactions (past and future imports/syncs honor it too). Use this to clean up transactions stuck in 'other'. Match is case-insensitive. Args: match (substring, e.g. 'menards'), category. Returns how many transactions were recategorized. Example: {match:'robinhood', category:'transfers'}.

List all merchant→category rules (most specific first) and re-apply them across the whole store, returning how many transactions changed. Call after editing rules to resync. No args.

Delete a category rule by its exact match string. Transactions it already recategorized keep their current category (rules don't store the original). Args: match. Returns whether a rule was removed.