Frappe MCP Server
Manage your Frappe benches and sites directly from Claude.ai through natural language, enabling operations like restarting services, listing sites, running Python scripts, and viewing logs.
README
Frappe Bench & Site Manager ā MCP Server
Manage your local Frappe benches and sites directly from Claude.ai (web + mobile).
Uses FastMCP over HTTP with an Ngrok static tunnel.
š± Claude.ai ā š Ngrok (permanent URL) ā š„ļø MCP Server (localhost:8000) ā š§ Frappe Benches
Tools Available
| Tool | Category | What it does |
|---|---|---|
bench_restart |
Bench Ops | Restart supervisor/bench services |
list_sites |
Site Management | List benches, sites, apps, status |
frappe_api |
DocType Data | Authenticated REST API calls |
bench_execute |
Console | Run Python in Frappe context |
get_logs |
Logs | Fetch and filter site/bench log files |
get_config_overview |
Config | Show configured benches (no secrets exposed) |
Prerequisites
- Python 3.10+
- One or more Frappe benches already set up and working locally
- ngrok account (free tier works)
Step 1: Install dependencies
git clone <this-repo>
cd frappe-mcp
pip install -r requirements.txt
Step 2: Configure
cp config.example.json config.json
Edit config.json:
{
"benches": [
{
"id": "bench1",
"label": "Main Dev Bench",
"path": "/home/youruser/frappe-bench",
"bench_cmd": "/home/youruser/frappe-bench/env/bin/bench"
}
],
"site_credentials": {
"mysite.localhost": {
"api_key": "YOUR_API_KEY",
"api_secret": "YOUR_API_SECRET",
"port": 8000
}
}
}
Important: config.json is gitignored ā never commit it.
Step 3: Get Frappe API Credentials
For each site you want to access via frappe_api or bench_execute:
- Open the Frappe site in your browser
- Go to Settings ā API Access
- Click Generate Keys (for your admin user)
- Copy
api_keyandapi_secretintoconfig.json ā site_credentials
Step 4: Start the MCP Server
python main.py
You should see:
š Frappe MCP Server starting...
Benches configured : 1
Sites with creds : 1
Listening on : http://0.0.0.0:8000
MCP endpoint : http://0.0.0.0:8000/mcp
Audit log : mcp_audit.log
Step 5: Setup Ngrok (one-time)
Install ngrok
macOS (Homebrew):
brew install ngrok/ngrok/ngrok
Linux:
curl -sSL https://ngrok-agent.s3.amazonaws.com/ngrok.asc | sudo tee /etc/apt/trusted.gpg.d/ngrok.asc >/dev/null
echo "deb https://ngrok-agent.s3.amazonaws.com buster main" | sudo tee /etc/apt/sources.list.d/ngrok.list
sudo apt update && sudo apt install ngrok
Or download directly: https://ngrok.com/download
Add your auth token
ngrok config add-authtoken YOUR_AUTHTOKEN
Get your token from: https://dashboard.ngrok.com/get-started/your-authtoken
Start tunnel with your free static domain
ngrok http 8000 --domain=yourname.ngrok-free.app
Get a free static domain: ngrok Dashboard ā Cloud Edge ā Domains ā New Domain
Your permanent MCP URL becomes:
https://yourname.ngrok-free.app/mcp
Step 6: Connect to Claude.ai
- Open claude.ai ā Settings ā Integrations
- Click Add Integration
- Enter URL:
https://yourname.ngrok-free.app/mcp - Click Add ā that's it
Works from your phone and laptop anywhere as long as your laptop is running.
Step 7: Auto-start on Boot (optional)
Linux (systemd)
Create /etc/systemd/system/frappe-mcp.service:
[Unit]
Description=Frappe MCP Server
After=network.target
[Service]
Type=simple
User=youruser
WorkingDirectory=/home/youruser/frappe-mcp
ExecStart=/usr/bin/python3 /home/youruser/frappe-mcp/main.py
Restart=on-failure
RestartSec=5
[Install]
WantedBy=multi-user.target
sudo systemctl daemon-reload
sudo systemctl enable frappe-mcp
sudo systemctl start frappe-mcp
macOS (launchd)
Create ~/Library/LaunchAgents/com.frappe.mcp.plist:
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>Label</key>
<string>com.frappe.mcp</string>
<key>ProgramArguments</key>
<array>
<string>/usr/bin/python3</string>
<string>/Users/youruser/frappe-mcp/main.py</string>
</array>
<key>WorkingDirectory</key>
<string>/Users/youruser/frappe-mcp</string>
<key>RunAtLoad</key>
<true/>
<key>KeepAlive</key>
<true/>
<key>StandardOutPath</key>
<string>/Users/youruser/frappe-mcp/mcp_stdout.log</string>
<key>StandardErrorPath</key>
<string>/Users/youruser/frappe-mcp/mcp_stderr.log</string>
</dict>
</plist>
launchctl load ~/Library/LaunchAgents/com.frappe.mcp.plist
Running Tests
python test_tools.py
No real Frappe bench needed ā all tools are tested with a mock config.
Security Notes
config.jsonis never committed (gitignored)- API keys are never exposed in tool responses (
get_config_overviewshows"configured"/"missing"only) - All subprocess calls use
shell=Falseā no shell injection possible bench_executehas a two-tier security scanner: hard blocks and soft warnings- Blocked patterns (configurable):
DROP,TRUNCATE,os.system,exec(,eval(, etc. - Rate limit: 30 requests/minute per IP (configurable)
- All tool calls are appended to
mcp_audit.log
Example Claude Prompts
List all my Frappe sites and their status.
How many NGO records are on site1.localhost in bench1?
Show me the last 50 error log lines for site1.localhost ā filter for "PermissionError".
Restart the web worker on bench1.
What's in the mGrant Settings module field for site1.localhost?
Show me all active Grants on site1.localhost with fields name, grant_name, grant_status.
File Structure
frappe-mcp/
āāā main.py # FastMCP server + tool registration
āāā config.py # Config loader + validator + singleton
āāā security.py # Input sanitizer, command blocker, rate limiter
āāā logger.py # Audit logger ā mcp_audit.log
āāā tools/
ā āāā bench_ops.py # bench_restart (+ Phase 2 stubs)
ā āāā site_manager.py # list_sites (+ Phase 2 stub)
ā āāā frappe_api.py # frappe_api (+ Phase 2 stub)
ā āāā executor.py # bench_execute
ā āāā log_reader.py # get_logs
āāā config.json # Your config (gitignored)
āāā config.example.json # Template (committed)
āāā requirements.txt
āāā .gitignore
āāā test_tools.py # Test suite (mock config, no bench needed)
āāā README.md
Troubleshooting
config.json not found
ā Run cp config.example.json config.json and fill in your bench paths.
Bench path '/home/...' does not exist
ā The path in config.json ā benches must be an existing directory.
bench command not found
ā Use the full path to the bench binary, e.g. /home/user/frappe-bench/env/bin/bench.
No credentials configured for site
ā Add api_key/api_secret for that site to config.json ā site_credentials.
Authentication failed (401)
ā Regenerate API keys in the Frappe site: Settings ā API Access.
Ngrok shows ERR_NGROK_3200
ā Your static domain may not be activated. Check ngrok Dashboard ā Domains.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.