nr-mcp
Lets AI assistants interact with Node-RED to read flows, search nodes, edit function code, deploy changes safely, and manage modules.
README
nr-mcp
<!-- mcp-name: io.github.Texan-NXTassist/nr-mcp -->
A Model Context Protocol (MCP) server that lets AI assistants interact with Node-RED — read flows, search nodes, edit function code, deploy changes safely, and manage modules.
Built to solve real problems: the existing Node-RED MCP implementations use PUT /flow/:id which reorders your tabs. nr-mcp uses the correct GET → POST /flows pattern with optimistic locking, so your tab order is always preserved.
Features
13 tools for complete Node-RED flow management:
| Tool | Description |
|---|---|
nr_get_flow_summary |
Overview of all tabs with node counts and groups |
nr_get_flow |
Get a single tab with all nodes — by name or ID |
nr_search_nodes |
Search nodes by name, type, or JavaScript code content |
nr_get_function_code |
Extract full JS code from function nodes (incl. init/finalize) |
nr_get_node_config |
Full config with computed upstream/downstream connections |
nr_get_flow_context |
Read flow-level context variables |
nr_safe_deploy |
Deploy changes with optimistic locking — never reorders tabs |
nr_create_nodes |
Batch-create nodes/groups in a single deploy |
nr_delete_nodes |
Batch-delete with automatic wire and group cleanup |
nr_inject |
Trigger inject nodes remotely to test flows |
nr_get_installed_modules |
List installed modules and available node types |
nr_install_module |
Install npm packages from the Node-RED registry |
nr_get_debug_output |
Read debug/error data from flow context |
Why not just use the existing MCP servers?
- Tab reorder bug: Other implementations use
PUT /flow/:idwhich silently reorders your tabs in Node-RED. nr-mcp uses the correctGET → POST /flowsfull-deploy pattern. - Optimistic locking: Every deploy checks the
revfield. If someone else deployed between your read and write, you get a clear conflict error instead of silent data loss. - Smart search: Search across node names, types, function code, templates, and actions in one call.
- Production-tested: Built and used daily for managing complex Node-RED installations (home automation, IoT, energy management).
Installation
Prerequisites
Install from PyPI
pip install nr-mcp
Install with uv (recommended)
uv tool install nr-mcp
Or from source:
git clone https://github.com/Texan-NXTassist/nr-mcp.git
cd nr-mcp
uv tool install .
This creates the nr-mcp command in ~/.local/bin/.
Configuration
Environment variables
| Variable | Required | Description |
|---|---|---|
NR_URL |
No | Node-RED URL (default: http://localhost:1880) |
NR_TOKEN |
No* | Bearer token for token-based auth |
NR_USER |
No* | Username for Basic Auth |
NR_PASS |
No* | Password for Basic Auth |
* At least one auth method is recommended. Auth is checked in order: token → basic auth → no auth.
Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"nr-mcp": {
"command": "nr-mcp",
"env": {
"NR_URL": "http://localhost:1880",
"NR_USER": "admin",
"NR_PASS": "your-password"
}
}
}
}
Tip: If you get a "working directory" error, create a wrapper script:
#!/bin/bash cd /tmp exec nr-mcp "$@"Then point
commandto the wrapper path.
Cursor / VS Code
Add to your MCP settings (.cursor/mcp.json or VS Code equivalent):
{
"mcpServers": {
"nr-mcp": {
"command": "nr-mcp",
"env": {
"NR_URL": "http://localhost:1880",
"NR_TOKEN": "your-access-token"
}
}
}
}
Usage examples
Once connected, you can ask your AI assistant things like:
- "Show me all Node-RED tabs and their node counts"
- "Search for nodes that contain 'mqtt' in their code"
- "Show me the function code for node abc123"
- "Update the function code in node xyz to add error handling"
- "Create an inject node and an HTTP request node on the Weather tab"
- "What modules are installed? Is node-red-dashboard available?"
- "Install node-red-contrib-influxdb"
- "Trigger the 'Test' inject node and check for errors"
Architecture
nr-mcp is a Python MCP server that communicates with Node-RED via its Admin API. It uses stdio transport (standard for MCP) and makes HTTP calls to your Node-RED instance.
AI Assistant ↔ MCP (stdio) ↔ nr-mcp ↔ HTTP ↔ Node-RED Admin API
Key design decisions
- GET→POST pattern: All deploys fetch full flows, modify in-place, then POST back. Never uses
PUT /flow/:idwhich reorders tabs. - Optimistic locking: Uses the Node-RED
revfield to detect concurrent modifications. - No caching: Every tool call fetches fresh data. Slightly slower, but always correct.
- Single retry on conflict: Deploy operations retry once on 409 Conflict.
For more details, see docs/ARCHITECTURE.md.
Node-RED setup
The Admin API must be enabled (it is by default). If you've restricted it, ensure these endpoints are accessible:
GET /flowsandPOST /flows— flow read/writeGET /context/flow/:id— flow contextPOST /inject/:id— inject triggerGET /nodesandPOST /nodes— module management
See Node-RED Admin API docs for auth configuration.
Contributing
Contributions welcome! Please open an issue first to discuss what you'd like to change.
git clone https://github.com/Texan-NXTassist/nr-mcp.git
cd nr-mcp
uv venv && source .venv/bin/activate
uv pip install -e .
License
MIT — see LICENSE.
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.