monzo-mcp
MCP server for the Monzo banking API with OAuth, local transaction cache, and spending analysis
Install / Use
claude mcp add partymola -- npx -y github:partymola/monzo-mcpIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
Finance & AccountingSupported Platforms
Our assessment of monzo-mcp
monzo-mcp scores 78/100 on our quality scale, 95th of 112 Finance & Accounting skills we index.
Its MCP Server is 9.1 KB long, well organised into 18 sections with 7 code examples: a thorough specification that gives an agent plenty to work with.
It has 3 GitHub stars, so there is little community track record yet; judge it on its content.
Maintenance, license and trust
- The repository was last updated 3 days ago, so monzo-mcp is actively maintained.
- Our last check on 2026-09-18 found the source still online.
- It is released under GPL-3.0, a copyleft license: you can use it, but modified versions you distribute must carry the same license.
- Its trust signals score 87/100, with 2 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-09-29. Automated pattern scan on 2026-09-29. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
monzo-mcp compared with similar skills
All 4 of these similar skills score higher than monzo-mcp; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| monzo-mcp (this skill)by partymola | 78 | 3 | 3d ago | MCP Server |
| Agent-Reachby Panniantong | 100 | 87.6k | 16d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 74.3k | today | CLAUDE.md |
| rufloby ruvnet | 100 | 73.7k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 47.2k | today | CLAUDE.md |
Frequently asked questions
- How do I install monzo-mcp?
- Run
claude mcp add partymola -- npx -y github:partymola/monzo-mcp. The install tabs above show the steps for each supported agent. - Which AI agents does monzo-mcp work with?
- It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
- Is monzo-mcp safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is GPL-3.0-licensed and scores 87/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is monzo-mcp still maintained?
- The repository was last updated 3 days ago, so monzo-mcp is actively maintained.
Skill content
View source on GitHubmonzo-mcp
MCP server for the Monzo banking API. Read-only access to accounts, balances, pots, transactions, and spending analysis - all through Claude Code or any MCP client.
Unlike other Monzo MCP implementations that use raw bearer tokens (which expire in 6 hours), this server handles full OAuth with automatic token refresh.
Features
- 7 read-only tools - no write operations, no money movement
- OAuth with auto-refresh - tokens refresh automatically, no manual regeneration
- Local transaction cache - SQLite database survives Monzo's 90-day SCA window
- Auto-sync on demand - the cache-reading tools run an incremental sync automatically if the cache wasn't synced today, so you rarely need to call
monzo_syncby hand - Spending analysis - category breakdowns, top merchants, month-over-month comparison
- Transaction search - search by merchant, payee (counterparty), description, or notes across cached history
- Counterparty details - bank transfers (faster payments, p2p, Bacs) are cached with the payee's name, sort code, and account number
Tools
| Tool | Description | Data source |
|------|-------------|-------------|
| monzo_list_accounts | List accounts with types and IDs | Live API |
| monzo_get_balance | Current balance and spend today | Live API |
| monzo_list_pots | Savings pots and balances | Live API |
| monzo_sync | Sync transactions to local cache | Live API -> SQLite |
| monzo_list_transactions | List/filter cached transactions | Local cache (auto-syncs if stale) |
| monzo_search_transactions | Search by merchant/payee/description/notes | Local cache (auto-syncs if stale) |
| monzo_spending | Spending analysis with category breakdown | Local cache (auto-syncs if stale) |
Prerequisites
- Python 3.13+
- uv (recommended) or pip
- A Monzo account with an OAuth client registered at developers.monzo.com
Installation
git clone https://github.com/partymola/monzo-mcp.git
cd monzo-mcp
uv venv --python 3.13 .venv
uv pip install -e .
This package is not on PyPI - the name belongs to an unrelated project. A container image is published instead; see Docker.
Setup
1. Register a Monzo OAuth client
Go to developers.monzo.com and create an OAuth client:
- Set the redirect URL to
http://localhost:6600/callback - Note your Client ID and Client Secret
2. Authenticate
monzo-mcp auth
This opens your browser for Monzo OAuth. After authorizing, approve the login in your Monzo app within 5 minutes for full transaction history access (Monzo's SCA window).
3. Register with Claude Code
claude mcp add -s user monzo -- /path/to/monzo-mcp/.venv/bin/monzo-mcp
4. First sync
In Claude Code, run monzo_sync to populate the local transaction cache. Do this immediately after auth to take advantage of the SCA window (up to 11 months of history).
Docker
Images are published to ghcr.io/partymola/monzo-mcp. Tags carry a v prefix (:vX.Y.Z), and :latest follows the most recent release.
The container needs a volume. Credentials and the transaction cache live under /data; with nothing mounted there, the container still starts and every tool reports that it is not configured, and anything you authorise is lost as soon as the container is replaced.
First register an OAuth client as described in Setup step 1 - auth prompts for the Client ID and secret, and there is no way to supply them later. The redirect URL is http://localhost:6600/callback, the same as a source install.
Authenticate once, into a named volume. Decide before you run this whether you want a bind mount instead - switching afterwards means authorising again:
docker volume create monzo-mcp-data
docker run --rm -it \
-v monzo-mcp-data:/data \
-p 127.0.0.1:6600:6600 \
ghcr.io/partymola/monzo-mcp:latest auth
The published port is needed only for this step, so the OAuth redirect can reach the container. Binding it to 127.0.0.1 keeps the callback listener off your network. No browser opens - the container has none - so copy the URL it prints. Then approve the login in your Monzo app within 5 minutes (see Monzo SCA window).
Then register the server, reusing the same volume:
claude mcp add -s user monzo -- \
docker run --rm -i -v monzo-mcp-data:/data ghcr.io/partymola/monzo-mcp:latest
-i is required - the server speaks JSON-RPC over stdin and stdout.
Sync straight away. The 11-month backfill closes 5 minutes after you approve in the Monzo app (see Monzo SCA window), and this route is longer than a source install - so call monzo_sync as soon as the server is registered. Miss it and you silently get 90 days instead, with no error.
To keep the files somewhere you can read them, use a bind mount instead of a named volume. The container runs as root and writes credentials owner-only, so without --user they end up root-owned:
mkdir -p ~/monzo-mcp-data/config
docker run --rm -it \
-v ~/monzo-mcp-data:/data \
--user $(id -u):$(id -g) \
-p 127.0.0.1:6600:6600 \
ghcr.io/partymola/monzo-mcp:latest auth
Pass the same -v and --user to the server command. Create the directory first - --user against a named volume fails, because the volume initialises root-owned from the image.
CLI
monzo-mcp Start the MCP server (stdio transport)
monzo-mcp auth Interactive OAuth setup (opens the browser)
monzo-mcp --version Print the installed package version
Configuration
All configuration is via environment variables (optional):
| Variable | Default | Description |
|----------|---------|-------------|
| MONZO_MCP_CONFIG_DIR | <package>/config/ | Directory for OAuth credentials and tokens |
| MONZO_MCP_DB_PATH | <package>/monzo.db | Path to SQLite transaction cache |
| MONZO_MCP_CALLBACK_HOST | localhost | Interface the auth callback server binds to. The redirect URI is unaffected |
The container image sets the first two under /data, because the package-relative defaults resolve into the interpreter's lib directory there, which cannot be mounted. It sets the third to 0.0.0.0, because a published port arrives on the container's bridge interface and a localhost bind refuses it.
Credential files (created by monzo-mcp auth):
config/monzo_client.json- OAuth client ID and secretconfig/monzo_tokens.json- Access and refresh tokens (auto-refreshed)
Monzo SCA window
Monzo's Strong Customer Authentication (SCA) limits transaction history access:
- Within 5 minutes of app approval: up to ~11 months of history
- After the window expires: only the last 90 days
The local SQLite cache preserves all synced transactions permanently, so run monzo_sync promptly after monzo-mcp auth.
To backfill a specific range, pass since to monzo_sync - an ISO date (2026-01-01) or datetime (2026-01-01T14:30:00Z). Reaching back more than ~90 days only works inside the SCA window; outside it, only the last 90 days are returned.
Older cached transactions gain fields added in newer versions (e.g. counterparty/payee details on bank transfers) only when re-fetched, which a post-auth full sync does for the history it re-pulls.
Security
- Zero write tools - cannot send money, move funds between pots, or modify transactions
- Monzo API itself cannot send money to external accounts
- Tokens stored as JSON files in the
config/directory (gitignored) - All API calls are GET requests with Bearer token auth
Troubleshooting
- "SCA required" or only 90 days of history - re-run
monzo-mcp authand approve the login in the Monzo app within 5 minutes, then sync straight away (see the SCA window above). - Token expired / no refresh - re-run
monzo-mcp authto re-authorise. - "No transaction data available" - the cache is empty; call
monzo_sync(or any cache-reading tool, which auto-syncs) once after authenticating. - Every tool reports "not configured" under Docker, or authentication does not survive a restart - nothing is mounted at
/data. See Docker; the same volume must be passed to theauthrun and to the server. authsits on "Waiting for callback..." after you approve - the callback was not delivered. Press Ctrl-C and run it again. Under Docker, check the port is published (-p 127.0.0.1:6600:6600).
Contributing
See CONTRIBUTING.md for development setup, the test workflow, and the pre-commit hook. Changes are tracked in CHANGELOG.md.
License
GPL-3.0-or-later. See LICENSE.
Related Skills
Agent-Reach
87.6kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
74.3kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
ruflo
73.7k🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
CowAgent
47.2kOpen-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
