mqtt
Lightweight pub/sub protocol for efficient IoT communication and telemetry
How do I install this agent skill?
npx skills add https://github.com/alphaonedev/openclaw-graph --skill mqttIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill facilitates MQTT protocol communication for IoT tasks using the standard paho-mqtt library and mosquitto CLI tools. It follows best practices by recommending environment variables for credentials. However, it introduces an attack surface for indirect prompt injection from external message payloads and potential command injection risks if subprocess calls are used with unsanitized inputs.
- Socketpass
No alerts
- Snykwarn
Risk: MEDIUM · No issues
- Runlayerpass
1 file scanned · No issues
What does this agent skill do?
mqtt
Purpose
This skill allows the AI to interact with MQTT, a lightweight pub/sub protocol, for efficient IoT communication. Use it to publish telemetry data or subscribe to device updates, enabling real-time interactions in constrained environments.
When to Use
Use this skill for scenarios requiring low-bandwidth, real-time messaging, such as sending sensor data from IoT devices, remote device control, or telemetry monitoring. Avoid it for high-throughput needs like file transfers, where HTTP is more suitable.
Key Capabilities
- Publish messages to topics with configurable Quality of Service (QoS) levels (0, 1, or 2).
- Subscribe to topics and handle incoming messages asynchronously.
- Support for retained messages, last will and testament, and wildcard subscriptions (e.g., "#" for all subtopics).
- Secure connections via TLS and authentication methods like username/password.
Usage Patterns
To use MQTT, first connect to a broker using a client library. Establish a connection with credentials, then publish or subscribe. For example, in a loop: connect, subscribe, process messages, and disconnect. Use environment variables for sensitive data, like $MQTT_BROKER_URL for the broker address. Always handle reconnections for unstable networks.
Common Commands/API
Use the Paho MQTT library for Python interactions. Import it as follows:
import paho.mqtt.client as mqtt
To connect to a broker:
- Command:
client.connect(broker_address, port=1883) - Example with auth:
client.username_pw_set(username='$MQTT_USERNAME', password='$MQTT_PASSWORD')
To publish a message:
- API:
client.publish(topic, payload, qos=1) - Example:
client.publish('sensors/temp', '22.5', qos=1)
To subscribe:
- API:
client.subscribe(topic, qos=1) - Example:
client.subscribe('commands/device1', qos=1)
Config format: Use a JSON file for settings, e.g., {"broker": "$MQTT_BROKER_URL", "topic": "sensors/temp"}. CLI tools like mosquitto_pub can be invoked via subprocess: subprocess.run(['mosquitto_pub', '-h', '$MQTT_BROKER_URL', '-t', 'sensors/temp', '-m', '22.5']).
Integration Notes
Integrate MQTT by wrapping it in your AI agent's code. Set environment variables for keys, e.g., export $MQTT_API_KEY for token-based auth. For multi-service setups, use a broker like HiveMQ or Mosquitto; connect via client.connect('$MQTT_BROKER_URL'). Ensure TLS for secure endpoints (e.g., client.tls_set()). If combining with other skills, like a database, publish data on success: e.g., after inserting into a DB, run client.publish('db/updates', json.dumps(data)).
Error Handling
Handle common errors explicitly. For connection issues, catch ConnectionRefusedError and retry with exponential backoff: e.g., time.sleep(5). For publish failures, check return codes (e.g., if client.publish() returns >0, log and retry). Subscription errors might involve invalid topics; validate with if '/' not in topic: raise ValueError('Invalid topic'). Use callbacks for message errors, like in on_message handler: if error: print('Error: ', error). Always clean up with client.disconnect() in a finally block.
Concrete Usage Examples
Example 1: Publish sensor data
To publish a temperature reading to an MQTT topic, first set env vars: export MQTT_BROKER_URL='broker.example.com'. Then, in code:
client = mqtt.Client()
client.connect(os.environ['MQTT_BROKER_URL'])
client.publish('sensors/temperature', '25.0', qos=1)
client.disconnect()
This sends a message to the broker for IoT dashboard updates.
Example 2: Subscribe to device commands To subscribe and react to commands, use:
def on_message(client, userdata, message):
print(f"Received: {message.payload.decode()}")
client = mqtt.Client()
client.on_message = on_message
client.connect(os.environ['MQTT_BROKER_URL'])
client.subscribe('commands/device1', qos=1)
client.loop_forever()
This keeps the client listening for commands to control an IoT device.
Graph Relationships
- Related to cluster: iot
- Related to tags: iot, mqtt
- Connections: Can integrate with other iot skills, such as for data processing or device management.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/alphaonedev/openclaw-graph/mqtt">View mqtt on skillZs</a>