An agent to run FlowFuse managed instances of Node-RED on devices
Repo FlowFuse/device-agent, read from GitHub 1 min ago. FlowFuse has not joined Take My Repo.
FlowFuse/device-agent is a public repo on GitHub by FlowFuse. An agent to run FlowFuse managed instances of Node-RED on devices. It declares the Apache-2.0 license on GitHub. It is written mainly in JavaScript, has 20 stars and last changed on 2026-11-19. Nobody has reviewed it on Take My Repo yet, which says nothing about its quality. Read from GitHub on 2026-11-21.
In the maintainer's words. Shown from GitHub; Take My Repo does not keep a copy.
This module provides an agent that runs Node-RED instances deployed from the FlowFuse platform.
Important
The v4 release of the device agent brings some potentially breaking changes:
latest docker tag now points at a Node.js 22 based containerMore information about these changes are available in the docker section.
For users that require Node.js v14 and v16 support the v2.x stream will work.
The Device Agent can be installed on most Linux distributions, Windows, and MacOS.
A one-line installer is available for the Device Agent. This will install a Node.js runtime and the device agent, and also set it up to run as a system service.
More details available in the FlowFuse Device Agent Installer documentation
If you already have Node.js installed, you can install the agent via npm.
The Device Agent is published to the public npm repository as @flowfuse/device-agent.
It can be installed as a global npm module. This will ensure the agent command is on the path.
Note: previous versions of the agent were published as @flowforge/flowforge-device-agent.
sudo npm install -g @flowfuse/device-agent
npm install -g @flowfuse/device-agent
We publish a Docker container for the Device Agent as flowfuse/device-agent on DockerHub.
Containers are built for a range of Node.js versions and tagged accordingly: latest-18, latest-20, latest-22 and latest-24.
With the Device Agent v4 release, the latest tag points to the latest-22 container.
When running with the container you will need to mount the device.yml obtained when Registering the device:
docker run --mount type=bind,src=/path/to/device.yml,target=/opt/flowfuse-device/device.yml -p 1880:1880 flowfuse/device-agent:latest
Alternatively, you can mount the entire configuration directory:
docker run --mount type=bind,src=/path/to/config/dir,target=/opt/flowfuse-device -p 1880:1880 flowfuse/device-agent:latest
Important
Breaking change in Device Agent v4 — the container now runs as a non-root user.
From 4.0.0 the image runs as the unprivileged flowfuse user (UID 2000 /
GID 2000) instead of root. If you bind-mount a host directory for state
(e.g. at /opt/flowfuse-device to persist device.yml, the project/
directory and the module cache), that directory must be writable by UID
2000, otherwise the agent will fail to start with a permissions error.
Before upgrading, change the ownership of the mounted directory on the host:
sudo chown -R 2000:2000 /path/to/config/dir
Alternatively, run the container as a UID that already owns the host directory (the agent only requires the state directory to be writable by the runtime user):
docker run --user $(id -u):$(id -g) ... flowfuse/device-agent:latest
If you build the image yourself, the user and group IDs can be customised with
the FF_UID and FF_GID build arguments.
The agent configuration is provided by a device.yml file within its working
directory.
By default the agent uses /opt/flowfuse-device or c:\opt\flowfuse-device as
its working directory. This can be overridden with the -d/--dir option.
For backwards compatibility with previous versions, the agent will use /opt/flowforge-device
if it is exists, unless overridden on the command-line.
NOTE: The device agent will attempt to create the working directory if it is not found, however if an error occurs, the device agent will exit and report a startup error.
sudo mkdir /opt/flowfuse-device
sudo chown -R $USER /opt/flowfuse-devicemkdir c:\opt\flowfuse-device
device.yml - for a single deviceWhen the device is registered on the FlowFuse platform, a group of configuration details are provided. These can be copied from the platform, downloaded directly as a yml file or pulled from the FlowFuse server using the command issued by the platform when you create a device. More details on this can be found in the FlowFuse documentation.
This file should exist in the working directory as device.yml.
A different config file can be specified with the -c/--config option.
The file must contain the following options (these are the ones provided by FlowFuse)
| Required options | Description |
|---|---|
deviceId |
The id for the device on the FlowFuse platform |
token |
Access Token to connect to the FF platform |
credentialSecret |
Key to decrypt the flow credentials |
forgeURL |
The base url of the FlowFuse platform |
To enable MQTT connectivity, the following options are required. They are provided by the platform if MQTT comms are enabled.
| MQTT options | Description |
|---|---|
brokerURL |
The url for the platform broker |
brokerUsername |
The username to connect with - device:<teamId>:<deviceId> |
brokerPassword |
The password to connect with |
The following options can be added:
| Extra options | Description |
|---|---|
interval |
How often, in seconds, the agent checks in with the platform. Default: 60s |
intervalJitter |
How much, in seconds, to vary the heartbeat +/- intervalJitter. Default: 10s |
moduleCache |
If the device can not access npmjs.org then use the node modules cache in module_cache directory. Default false |
verbose |
Enables verbose logging |
jsonLogging |
Enabled JSON format logs |
Used to pass options to the Node.js process running Node-RED
nodeOptions | An array of Node.js command line options. Default is single entry of --max-old-space-size=512
nodeOptions:
- '--max-old-space-size=256'
- '--use-openssl-ca'
The following options are passed through to Node-RED:
| Node-RED options | Description |
|---|---|
port |
The port to listen on. Default: 1880 |
https |
Enable HTTPS. See below for details |
httpStatic |
Enable serving of static content from a local path |
httpNodeAuth |
If set, any endpoints created in Node-RED flows will require this username and password to access them. Default: false |
localAuth |
When configured, local login is possible |
https configurationThe https configuration option can be used to enable HTTPS within Node-RED. The values
are passed through to the Node-RED https setting.
The ca, key and cert properties can be used to provide custom certificates and keys.
The values should be set to the contents of the certificate/key.
Alternatively, the properties caPath, keyPath and certPath can be used instead
to provide absolute paths to files containing the certificates/keys.
https: keyPath: /opt/flowfuse-device/certs/key.pem certPath: /opt/flowfuse-device/certs/cert.pem caPath: /opt/flowfuse-device/certs/ca.pem
httpStatic configurationThis option can be used to serve content from a local directory.
If set to a path, the files in that directory will be served relative to /.
httpStatic: /opt/flowfuse-device/static-content
It is also possible to configure it with a list of directories and the corresponding path they should be served from.
httpStatic: - path: /opt/flowfuse-device/static-content/images root: /images - path: /opt/flowfuse-device/static-content/js root: /js
httpNodeAuth configurationThis option can be used to apply basic auth to HTTP endpoints created in Node-RED. This will also protect node-red-dashboard. Full details can be found in HTTP Node security documentation.
Example:
httpNodeAuth: user: user pass: $2a$08$zZWtXTja0fB1pzD4sHCMyOCMYz2Z6dNbM6tl8sJogENOMcxWV9DN.
localAuth configurationThe recommended way to securely access the Node-RED editor is via FlowFuse platform.
However, if you want to enable local login within Node-RED, you can use the localAuth configuration.
Example:
localAuth: user: user pass: $2a$08$zZWtXTja0fB1pzD4sHCMyOCMYz2Z6dNbM6tl8sJogENOMcxWV9DN.
NOTE: localAuth.enabled can be set to false to disable local login without the need to remove the user and pass properties.
The editor is located on the path /device-editor e.g. http://localhost:1880/device-editor
device.yml - for provisioningWhen a device should be auto registered on the FlowFuse platform, a group of provisioning configuration details are required. These are generated for you in FlowFuse Team Settings under the Devices tab when you create a provisioning token. These can be copied from the platform, or downloaded directly as a yml file.
This file should be copied into the working directory as device.yml.
A different config file can be specified with the -c/--config option.
The file must contain the following options (these are the ones provided by FlowFuse)
| Required options | Description |
|---|---|
provisioningName |
The name of the token |
provisioningTeam |
The team this device will be registered to |
provisioningToken |
Provisioning Token to connect to the FF platform |
forgeURL |
The base url of the FlowFuse platform |
deviceName |
Device Name (optional) |
The following options can be added:
| Extra options | Description |
|---|---|
interval |
How often, in seconds, the agent checks in with the platform. Default: 60s |
intervalJitter |
How much, in seconds, to vary the heartbeat +/- intervalJitter. Default: 10s |
verbose |
Enables verbose logging |
jsonLogging |
Enabled JSON format logs |
If the agent was installed as a global npm module, the command
flowfuse-device-agent will be on the path.
If the default working directory and config file are being used, then the agent can be started with:
$ flowfuse-device-agent
For information about the available command-line arguments, run with -h:
Options
-c, --config file Device configuration file. Default: device.yml
-d, --dir dir Where the agent should store its state. Default: /opt/flowfuse-device
-i, --interval secs
-p, --port number
-m, --moduleCache Use local npm module cache rather than install
--node-options Adds Node.js command line arguments, can be specified multiple times. Must use = e.g. --node-options="--max-old-space-size=256"
Web UI Options
-w, --ui Start the Web UI Server (optional, does not run by default)
--ui-host string Web UI server host. Default: (0.0.0.0) (listen on all interfaces)
--ui-port number Web UI server port. Default: 1879
--ui-user string Web UI username. Required if --ui is specified
--ui-pass string Web UI password. Required if --ui is specified
--ui-runtime mins Time the Web UI server is permitted to run. Default: 10
Setup command
-o, --otc string Setup device using a one time code
--otc-no-start Do not start the agent after setup
--otc-no-import Do not ask to import Node-RED flows during setup
-u, --ff-url url URL of FlowFuse. Required for setup
Global Options
-h, --help print out helpful usage information
--version print out version information
-v, --verbose turn on debugging output
-j, --json-logging print logs in JSON format
By default the Device Agent will try and download the correct version of Node-RED and any nodes required to run the Snapshot that is assigned to run on the device.
If the device is being run on an offline network or security policies prevent the Device Agent from connecting to npmjs.org then it can be configured to use a pre-cached set of modules.
You can enable this mode by adding -m to the command line adding moduleCache: true
to the device.yml file. This will cause the Device Agent to load the modules from the
module_cache directory in the Device Agents Configuration directory as described above.
By default this will be /opt/flowfuse-device/module_cache.
To create a suitable module cache, the device must be assigned to an instance. You will need to install the modules on a local device with access to npmjs.org, ensuring you use the same OS and Architecture as your target device, and then copy the modules on to your device.
package.json file.npm install to install the modules. This will create a node_modules directory.module_cache inside the Device Agent Configuration directory.node_modules directory from your local device to the target device so that it is under the module_cache directory.An example service file is provided here.
The Device Agents Web UI is provided to enable the user to download a device configuration or a provisioning configuration file. This is an optional feature and is not enabled by default.
To enable the UI, use the -w/--ui option. This will start a web server on
the specified host and port. The default host is 0.0.0.0 and the default port is 1879.
When enabling the UI, a username and password must be provided with the
--ui-user and --ui-pass options.
The UI will only be available for the duration specified by the --ui-runtime. By
default this is 10 minutes. After this time, the web server will be disabled. The
application must be restarted to re-enable the UI. You can set this to 0 to
disable the timeout. This is not recommended.
The following scripts are available:
npm start - Start the agentnpm run dev - Build the agent and watch for changesnpm run lint - Run eslintnpm run lint:fix - Run eslint and fix any issuesnpm run test - Run all unit testsnpm run test:lib - Run the unit tests for the libnpm run test:frontend - Run the unit tests for the frontendIn this project, the Release Please is used to automatically determine the next release version based on the commit messages in the codebase.
By using the Conventional Commits, the project adheres to a standardized format for commit messages, which Release Please uses to determine whether the next release should be a major, minor, or patch release.
The Prepare release GitHub Action workflow:
package.json and creates a new release on GitHub with the appropriate changelogThe Lint Pull Request Title GitHub Action workflow:
amannn/action-semantic-pull-request action to validate that pull request titles follow the Conventional Commits formatThe Conventional Commits preset expects pull request titles to be in the following format:
<type>(<scope>): <subject>
feat: A new feature (triggers a minor version bump).fix: A bug fix (triggers a patch version bump).perf: A code change that improves performance (triggers a patch version bump).refactor: A code change that neither fixes a bug nor adds a feature (does not trigger a release unless it's accompanied by a BREAKING CHANGE).docs: Documentation-only changes (does not trigger a release).chore: Changes to the build process or auxiliary tools and libraries (does not trigger a release).To indicate a breaking change, the exclamation mark ! should be used immediately after the type/scope:
feat!:,fix!:refactor!:FlowFuse/device-agent is a public repo on GitHub by FlowFuse. An agent to run FlowFuse managed instances of Node-RED on devices.
It declares the Apache-2.0 license on GitHub.
Its maker has not tagged a system, so it is unknown here. Its README says what it needs.
Open "Take it" on its page: Take My Repo writes an install plan for your machine and hands it to your AI agent (Claude Code, Codex or Cursor), which asks you before each step. Take My Repo has not tested it on any machine.
Its last change on GitHub was on 2026-11-19. Take My Repo does not judge from that whether it is maintained.
Nobody has reviewed it on Take My Repo yet, which says nothing about its quality.

A Node-RED node that adds "Ask ChatGPT" option to a duplicate of the built-in function node.
This page is also available as structured data, with the same sources and dates.
/api/v1/resources/gh/FlowFuse/device-agentTake My Repo has not tested this on your machine. Requirements come from the project's own documentation.