About Foxmayn Frappe CLI
Foxmayn Frappe CLI provides a command-line tool, MCP server, and desktop app for Frappe and ERPNext sites. It is for site administrators and AI-client workflows, with controls such as per-site policies, confirmations, and local audit logs.
Key features
- 01Read and change documents, reports, and workflows
- 02Manage files and permissions through the REST API
- 03Output JSON, NDJSON, CSV, and other formats
- 04Set per-site tool and DocType policies
- 05Use confirmations and dry runs for writes
- 06Keep a local audit log of tool calls
Use cases
- 01Query or export site records from the CLI
- 02Connect AI clients to ERPNext data
- 03Approve document or workflow changes before execution
- 04Restrict tools and DocTypes available on each site
MCP support
`ffc mcp` exposes Frappe and ERPNext site tools to AI clients including Claude Desktop, Claude Code, Cursor, VS Code, and Codex. It supports read-only mode, per-site policies, change confirmations, and audit logging.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEffc — Foxmayn Frappe CLI
A command line, an MCP server and a desktop app for Frappe and ERPNext sites.
Not an official Frappe project. ffc and Foxmayn Frappe Desktop are made independently by Foxmayn and are not affiliated with or endorsed by Frappe Technologies.
- ffc works with documents, reports, workflows, files and permissions over the REST API, with output made for scripts (JSON, NDJSON, CSV,
--jq). ffc mcplets AI assistants (Claude Desktop, Claude Code, Cursor, VS Code, Codex) use your sites, with per-site limits, confirmations and an audit log.- Foxmayn Frappe Desktop (Windows and macOS) adds sites, chats with them through its Assistant and connects AI apps, without the terminal.
A real run of the Desktop Assistant (Claude Opus 5.5) on a demo ERPNext site, waits shown at 4× speed. With "Ask before changes" on, nothing is written until you approve the card.
Safety
An AI assistant connected to an ERP should not be able to surprise you. ffc and the desktop app act with your own Frappe user's permissions and only ever narrow them:
- Read-only by default in the desktop app, and one flag (
--read-only) or one config key for any AI client. - Per-site policy: allow or deny tools, DocTypes and server methods. Users, permissions, scripts and settings stay read-only unless you list them.
- Approvals: ffc asks before deletes, cancels, workflow actions and shares, through any AI client that can ask (
confirm: alwaysrefuses them in clients that cannot). In the desktop app's "Ask before changes" mode every change waits on a card that shows the fields before and after, and an update is refused if the document changed on the site since you saw it. - Audit log: every tool call, allowed or refused, is one JSON line on your machine.
- Dry runs: CLI commands that write accept
--dry-runand print what they would send.
Details and limits: MCP safety · Desktop approvals.
Install
Linux and macOS
curl -fsSL https://raw.githubusercontent.com/nasroykh/foxmayn_frappe_cli/main/install.sh | sh
Windows (PowerShell or cmd.exe)
powershell -ExecutionPolicy Bypass -Command "irm https://raw.githubusercontent.com/nasroykh/foxmayn_frappe_cli/main/install.ps1 | iex"
Or with a package manager: brew install nasroykh/tap/ffc (macOS, Linux), or scoop bucket add foxmayn https://github.com/nasroykh/scoop-bucket then scoop install ffc (Windows).
Both scripts download the latest release for your platform and check its SHA-256 checksum. Manual downloads, building from source, signature checks, updating (ffc update) and uninstalling: Installation.
Desktop app (beta): builds are not code-signed yet (help fund signing), so a browser download shows a Windows SmartScreen or macOS Gatekeeper warning. The install scripts download the release themselves and check its SHA-256 checksum (and its release signature on macOS) before installing:
# macOS
curl -fsSL https://raw.githubusercontent.com/nasroykh/foxmayn_frappe_cli/main/install-desktop.sh | sh
# Windows (PowerShell or cmd.exe)
powershell -ExecutionPolicy Bypass -Command "irm https://raw.githubusercontent.com/nasroykh/foxmayn_frappe_cli/main/install-desktop.ps1 | iex"
Or download the desktop-v… release from the Releases page; see Install the desktop app for the first-launch steps.
Quickstart
ffc init # add your first site (browser sign-in, API key or password)
ffc whoami # who you are on the site, roles, Frappe version
ffc list-docs -d Customer -f name,customer_name -l 5
ffc get-doc -d "Sales Invoice" -n ACC-SINV-2026-00001 --json
ffc list-docs -d ToDo --filters '{"status":"Open"}' --output csv > todos.csv
Connect an AI assistant to your sites:
ffc mcp install --client claude-desktop # or claude-code, cursor, vscode, codex; --read-only allows reads only
More in the Quickstart.
Documentation
| Topic | Where |
|---|---|
| Install, first site, sign-in methods, config file and env vars | Installation · Authentication · Configuration |
| Every command, global flags, output formats, exit codes, dry runs | CLI reference |
| MCP server: setup per client, running it, tools, safety and policy | MCP |
| Desktop app: install, sites, the Assistant, connecting apps, updates | Desktop |
| Release signing, where credentials live | Security |
| Problems and questions | Troubleshooting · FAQ |
| Building, testing, releasing, the desktop dev loop | Development |
The full index is docs/README.md.
For AI agents
skills/ holds agent skills for using ffc (start with ffc-core; then ffc-setup, ffc-mcp, ffc-bulk-lifecycle, ffc-reports-api, ffc-files-collab) and for working on this repository (ffc-dev and the ffc-dev-* skills). make skills-init links them into .claude/, .cursor/ and .agent/.
Support
ffc is free and stays free. If it saves you time, you can support its development on Ko-fi. The first goal is code signing, so the desktop app installs on Windows and macOS without security warnings; then test servers and development tools. Already use Wise? You can also send through Wise.
Contributing
Issues and pull requests are welcome. Start with Development: make build, make test, make lint. Every new command or MCP tool ships with tests against the fake Frappe site in internal/frappetest.
Credits
ffc builds on these open-source projects, with thanks to their authors:
- CLI and MCP server: Cobra and pflag, Charm (Lip Gloss, Huh, Bubbles, colorprofile), Resty, gojq, mcp-go, hujson, go-yaml, go-isatty and the Go x/sys packages.
- Desktop app: Wails, React, Base UI, shadcn/ui, Tailwind CSS, Tabler Icons, Nunito Sans (via Fontsource), cmdk, Embla Carousel, Vite and the other packages in
desktop/frontend/package.json. - Release tooling: GoReleaser and NSIS.
- Frappe Framework and ERPNext, the open-source platforms this project talks to. Frappe and ERPNext are trademarks of their respective owners, used here only to describe compatibility.
License
MIT © 2026 Nas (Foxmayn)
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.