Analysis updated 2026-05-18
Scan three months of bank CSVs to find forgotten subscriptions you are still being charged for, review each one, and approve a cancellation email.
Dispute a duplicate charge by letting Moxie draft the dispute email with the scanned receipt attached, then approve and send it from your own inbox.
Upload bank statements and run moxie scan to get a plain-language list of potential money leaks before deciding which ones to act on.
| jacobbrooke1/moxie | a-bissell/unleash-lite | abhiinnovates/whatsapp-hr-assistant | |
|---|---|---|---|
| Stars | 1 | 1 | 1 |
| Language | Python | Python | Python |
| Setup difficulty | moderate | hard | hard |
| Complexity | 2/5 | 4/5 | 3/5 |
| Audience | general | researcher | developer |
Figures from each repo's GitHub metadata at analysis time.
Pending independent security review, keep MOXIE_LIVE=false until you have read the code. Requires an AI API key or a local Ollama model.
Moxie is an open-source Python program that scans your bank transactions, finds money you are probably losing, and helps you get it back. It looks for zombie subscriptions you forgot about, duplicate charges, and refunds that were promised but never arrived. It is designed to sit between two extremes: tools that only track your spending and services that act automatically without asking. When Moxie finds a problem, it writes a draft action, such as a cancellation email or a formal dispute, and shows it to you before anything is sent. You can edit the draft, approve it, or skip it. Nothing sends unless you explicitly say yes and have enabled live mode. By default the tool produces only drafts, so exploring it with real bank data carries no risk of unintended actions. The program works with bank statement CSV or PDF files you export yourself, or a read-only bank connection through aggregators like Plaid or TrueLayer. There is also a receipt vault where you scan a photo of a receipt or connect your email inbox to pull receipts in automatically. When you dispute a charge, the relevant receipt is attached as evidence. All actions are recorded in a hash-chained audit log that you can verify later to confirm it has not been modified. A community skill library holds reusable instructions for cancelling or disputing with specific companies, each carrying a tracked success rate. The program runs on your machine with your own AI API key (Claude or a locally running Ollama model). It includes a web dashboard and a command-line interface. The project is labeled as early scaffold and is pending an independent security review, so it should be used with caution with live financial data. It does not move money: payments and transfers are explicitly out of scope. The project is licensed under MIT.
A Python AI agent that scans bank transactions for zombie subscriptions and wrong charges, drafts cancellations and disputes, and always asks your permission before sending anything.
Mainly Python. The stack also includes Python, CLI, Tesseract OCR.
MIT license, meaning you can use, modify, and distribute it freely for any purpose including commercial use.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly general.
This repo across BitVibe Labs
Verify against the repo before relying on details.