{"id":51617828,"url":"https://github.com/djspiceroute/money-map","last_synced_at":"2026-07-12T15:30:56.672Z","repository":{"id":368727450,"uuid":"1285680860","full_name":"djspiceroute/money-map","owner":"djspiceroute","description":"An open Agent Skills + stdlib-Python toolkit that turns messy bank/app financial exports into one clean, de-duplicated, categorized dataset and a private local dashboard. Works with any AI coding agent — or standalone. Runs entirely locally.","archived":false,"fork":false,"pushed_at":"2026-07-01T23:22:50.000Z","size":63,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-07-02T00:26:19.335Z","etag":null,"topics":["agent-skills","ai-agents","budgeting","csv","dashboard","fintech","net-worth","personal-finance","plaid","python","transactions"],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/djspiceroute.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":"ROADMAP.md","authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-07-01T03:42:58.000Z","updated_at":"2026-07-01T23:22:55.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/djspiceroute/money-map","commit_stats":null,"previous_names":["djspiceroute/money-map"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/djspiceroute/money-map","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/djspiceroute%2Fmoney-map","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/djspiceroute%2Fmoney-map/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/djspiceroute%2Fmoney-map/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/djspiceroute%2Fmoney-map/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/djspiceroute","download_url":"https://codeload.github.com/djspiceroute/money-map/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/djspiceroute%2Fmoney-map/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":35395749,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-07-12T02:00:06.386Z","response_time":87,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agent-skills","ai-agents","budgeting","csv","dashboard","fintech","net-worth","personal-finance","plaid","python","transactions"],"created_at":"2026-07-12T15:30:55.939Z","updated_at":"2026-07-12T15:30:56.666Z","avatar_url":"https://github.com/djspiceroute.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# money-map\n\n**Turn messy financial exports from every bank and app into one clean, de-duplicated, categorized dataset — and a private local dashboard.**\n\nmoney-map is an **agent skill** paired with a set of small, dependency-free Python scripts. Point it at a folder of CSV exports (or a live Plaid connection) and it normalizes wildly different formats into one canonical schema, conservatively merges and de-duplicates overlapping sources, categorizes spending with editable rules, detects recurring subscriptions and bills, and builds a self-contained HTML dashboard with spending, net-worth, debt, and trend insights.\n\nIt's built on the open [Agent Skills](https://agentskills.io/specification) format, so it isn't tied to any one assistant — it works with any AI coding agent that reads that format (Claude, Codex, Copilot CLI, Gemini CLI, …), or standalone (the pipeline scripts are just Python — see [what's implemented today](#whats-implemented-today)).\n\n\u003e 🔒 **Everything runs locally.** No data is uploaded anywhere. The scripts are Python-stdlib-only, and the dashboard is a single offline HTML file. Your financial data never leaves your machine.\n\n---\n\n## Why\n\nFinancial data is scattered across banks, credit cards, and finance apps — each exporting a different CSV shape, with different column names, date formats, and **sign conventions** (some use negative for spending, some positive, some split debit/credit into two columns). Stitching them together by hand is tedious and error-prone, and naively merging two sources double-counts every overlapping transaction. money-map handles the plumbing so you get one trustworthy ledger and a dashboard you can actually reason about.\n\n## What you get\n\n- **One canonical dataset** from many sources (built-in adapters for Apple Card, Copilot, Mint, Monarch, YNAB, Plaid, and two generic bank formats — adding a new bank is a single dict).\n- **Conservative de-duplication** that *flags* rather than deletes, separating high-confidence exact duplicates from cross-source matches that might really be transfer legs.\n- **Editable, code-free enrichment** — category rules, account aliases, and friendly recurring labels all live in small CSVs you can correct and rebuild.\n- **Automatic recurring/subscription detection** from amount-repetition over time (catches subs no rule knows about, and flags \"possibly cancelled\" ones).\n- **A private dashboard** — Overview, Net Worth, Spending, Recurring/Subscriptions, Accounts, Debt \u0026 Bills, Transactions, and multi-year Analytics.\n\n### What's implemented today\n\nmoney-map is delivered as **runnable scripts for the data pipeline** plus reference docs that explain the method behind each stage:\n\n| Stage | Status |\n|---|---|\n| Normalize (ingest + adapters) | ✅ script — `ingest.py` / `adapters.py` |\n| Merge \u0026 de-duplicate | ✅ script — `dedup.py` |\n| Plaid / Copilot conversion | ✅ scripts — `plaid_to_csv.py`, `copilot_to_csv.py` |\n| Categorize \u0026 enrich (category rules, recurring detection, account aliases) | ✅ script — `enrich.py` |\n| Dashboard | ✅ script — `build_dashboard.py` |\n\nSo **standalone (no AI), the scripts can produce a clean, normalized, de-duplicated, enriched dataset and a local offline dashboard.** The reference docs remain the source of truth for how each stage works and how to extend it.\n\n## How it works\n\n```\n  exports / pulls  -\u003e  NORMALIZE  -\u003e  MERGE + DEDUPE  -\u003e  CATEGORIZE/ENRICH  -\u003e  DASHBOARD\n  (any format)         (one schema)   (flag duplicates)   (rules + detection)    (insights)\n```\n\nCanonical schema (one row per transaction):\n\n```\ndate, description, amount (negative = money out), account, institution, category, subcategory, source\n```\n\n## Quickstart (try it on the bundled sample data)\n\nThe repo ships with synthetic sample exports so you can see the pipeline run end-to-end in seconds. Requires Python 3 (stdlib only — nothing to install).\n\n```bash\ncd scripts\n\n# 1. See which adapter each file matches (no output written)\npython3 ingest.py --in ../assets/sample_data --inspect\n\n# 2. Normalize every export into one canonical CSV\npython3 ingest.py --in ../assets/sample_data --out canonical_transactions.csv\n\n# 3. Audit for duplicates (flags, never deletes)\npython3 dedup.py --in canonical_transactions.csv --audit-only\n```\n\nYou'll see the Apple Card and generic-bank formats auto-detected, purchases normalized to negative amounts, and the cross-source duplicate (a Whole Foods charge appearing on two accounts) flagged for review. To build the master file after reviewing the audit:\n\n```bash\npython3 dedup.py --in canonical_transactions.csv --out master_transactions.csv --apply\npython3 enrich.py --in master_transactions.csv --out enriched_transactions.csv\npython3 build_dashboard.py --in enriched_transactions.csv --out \"Finance Dashboard.html\"\n```\n\n\u003e The sample data is entirely fictional (`ACME CORP`, `Example Bank`, fake store numbers). Replace `../assets/sample_data` with a folder of your own exports to use it for real.\n\n## Using it with an AI agent\n\nmoney-map is built to be **driven by an AI coding agent** for a non-technical user: you drop exports in a folder and say *\"import my transactions\"* or *\"build me a finance dashboard,\"* and the agent runs the pipeline, picks sensible defaults, and asks short plain-language questions only when something needs your judgment. [`SKILL.md`](SKILL.md) is the playbook the agent follows.\n\nBecause it uses the open [Agent Skills](https://agentskills.io/specification) format (a `SKILL.md` plus supporting files), you install it by dropping the `money-map/` folder into your agent's skills directory:\n\n| Agent | Where it goes |\n|---|---|\n| **Claude Code / Claude Desktop** | `~/.claude/skills/money-map/` |\n| **OpenAI Codex CLI** | its skills dir, or the cross-runtime `~/.agents/skills/money-map/` |\n| **GitHub Copilot CLI** | auto-discovered from installed skills |\n| **Gemini CLI** | activates via its skill mechanism |\n\n```bash\n# example: Claude Code\ngit clone https://github.com/djspiceroute/money-map.git ~/.claude/skills/money-map\n```\n\nOnce installed, the skill activates automatically when you ask about importing transactions, combining statements, tracking net worth, finding subscriptions, or building a budgeting/finance dashboard.\n\n**No skills support?** In a plain chatbot, custom GPT, or project workspace, paste [`SKILL.md`](SKILL.md) in as instructions/context and run the scripts yourself — same playbook. The only hard requirement is that the agent (or you) can **access the filesystem and run Python**; a chat with no code execution can advise on the data but can't run the pipeline.\n\n**No AI at all?** The scripts are plain stdlib Python — run the [Quickstart](#quickstart-try-it-on-the-bundled-sample-data) to get a clean, de-duplicated, enriched dataset and an offline dashboard.\n\n## Supported sources\n\n| Source | How you export it |\n|---|---|\n| **Apple Card** | card.apple.com → Card Balance → Statements → Export Transactions (CSV) |\n| **Copilot Money** | Settings → Export Data (CSV), or pull live via the community CLI |\n| **Mint / Monarch / YNAB** | each app's Transactions → Export CSV |\n| **Plaid** (live balances, liabilities, transactions) | official Plaid CLI → `plaid_to_csv.py` — see [`references/plaid-setup.md`](references/plaid-setup.md) |\n| **Any bank** | a plain CSV (single Amount column, or split Debit/Credit) auto-detects |\n\nAdding a bank or app that isn't recognized is one dict in [`scripts/adapters.py`](scripts/adapters.py) — see [`references/source-adapters.md`](references/source-adapters.md).\n\n## Configuration (all code-free, editable CSVs)\n\nCorrect any mistake by editing a CSV and rebuilding — never a code change. Templates are in [`assets/`](assets/):\n\n| File | Purpose |\n|---|---|\n| `merchant_categories.example.csv` | substring → category rules (applied to spending only) |\n| `account_aliases.example.csv` | merge generic labels like `CREDIT CARD` into real account names |\n| `recurring_labels.example.csv` | rename cryptic bank descriptors to friendly names |\n| `manual_balances.example.csv` | add any account (investments, mortgage, property) to net worth |\n| `sources.example.json` | pin adapters/accounts when auto-detection needs help |\n\nSee [`references/enrichment.md`](references/enrichment.md) for the full method.\n\n## Repository layout\n\n```\nmoney-map/\n├── SKILL.md                 # the skill the agent follows (start here)\n├── ROADMAP.md               # feature ideas + credit to the OSS apps that inspired them\n├── references/              # deep-dive docs, loaded on demand\n│   ├── source-adapters.md   # canonical schema + adding new sources\n│   ├── enrichment.md        # categories, aliases, recurring detection, net worth\n│   ├── dashboard.md         # dashboard screens + build architecture\n│   ├── plaid-setup.md       # optional live bank connection via Plaid\n│   └── refresh.md           # keeping the data current\n├── scripts/                 # dependency-free Python (stdlib only)\n│   ├── ingest.py            # normalize a folder of exports → canonical CSV\n│   ├── adapters.py          # source adapters + auto-detection\n│   ├── dedup.py             # conservative merge + duplicate audit\n│   ├── enrich.py            # category rules, account aliases, recurring detection\n│   ├── build_dashboard.py   # offline single-file HTML dashboard\n│   ├── plaid_to_csv.py      # Plaid JSON dump → canonical CSV\n│   └── copilot_to_csv.py    # Copilot CLI JSON → canonical CSV\n└── assets/                  # editable config templates + synthetic sample data\n```\n\n## Privacy\n\n- **Local-only.** Scripts use the Python standard library only; the dashboard is one offline HTML file. Nothing is sent over the network (Plaid, if you opt in, runs via the official CLI on your machine).\n- **The repo contains no real data** — only synthetic samples and `.example` config templates.\n- **The [`.gitignore`](.gitignore) blocks real data by default.** All `*.csv`/`*.json` are ignored except the bundled examples, so if you run the pipeline inside a clone, your own exports and generated dashboards won't be committed. Don't put real exports in `assets/`.\n\n## Roadmap\n\nSee [ROADMAP.md](ROADMAP.md) for the **Now / Next / Later** roadmap: *Now* finishes the dashboard to its documented spec (net-worth trend, ranges, richer analytics), *Next* adds budgeting / savings-rate / bill status, and *Later* covers investments and FIRE — each item credited to the open-source app that inspired it.\n\n## Contributing\n\nAdapters for more banks/apps are the most useful contributions — most are a single dict in `scripts/adapters.py` plus a note in `references/source-adapters.md`. Please keep the scripts dependency-free (stdlib only) and never commit real financial data.\n\n## Credits \u0026 acknowledgements\n\nmoney-map stands on other people's work:\n\n**Integrations**\n- [**Plaid**](https://plaid.com) and its official [Plaid CLI](https://plaid.com/docs/resources/cli/) — the optional live bank connection (balances, liabilities, transactions), running locally.\n- [**copilot-money-cli**](https://github.com/JaviSoto/copilot-money-cli) by [**JaviSoto**](https://github.com/JaviSoto) — the unofficial community CLI used to pull Copilot Money data without a manual export.\n\n**Export formats adapted** — thanks to the apps whose CSV exports the built-in adapters read: Apple Card, [Copilot Money](https://copilot.money), Mint, [Monarch Money](https://www.monarchmoney.com), and [YNAB](https://www.ynab.com).\n\n**Feature \u0026 metric inspiration** — the roadmap draws ideas from these excellent open-source finance projects: [Wallos](https://github.com/ellite/Wallos), [Sure](https://github.com/we-promise/sure), [Firefly III](https://github.com/firefly-iii/firefly-iii), [Actual](https://github.com/actualbudget/actual), [Maybe](https://github.com/maybe-finance/maybe), [Paisa](https://github.com/ananthakumaran/paisa), and [Mintable](https://github.com/kevinschaich/mintable).\n\n**Built with** [Claude Code](https://claude.com/claude-code) — the skill, pipeline, and docs were designed and iterated with Claude.\n\n## License\n\n[MIT](LICENSE) — © 2026 djspiceroute.\n\n*Not affiliated with Plaid, Copilot Money, Apple, Mint, Monarch, YNAB, or any project listed above. This is a personal-use tool, not financial advice.*\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdjspiceroute%2Fmoney-map","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fdjspiceroute%2Fmoney-map","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdjspiceroute%2Fmoney-map/lists"}