The Agent Skills Directory

[SAFE]: The skill consists of C# implementation guides and architectural patterns for using MQTTnet in IoT plugins. Analysis of the provided code samples confirms they follow standard software development practices for asynchronous communication and background service management.
[INDIRECT_PROMPT_INJECTION]: The skill defines patterns for ingesting and processing data from external MQTT brokers, which is the primary purpose of the library. The documentation includes security best practices for handling this data, such as using JsonSerializer.Deserialize for configuration payloads and int.TryParse for numeric state values, which prevents basic injection or malformed data attacks.
[COMMAND_EXECUTION]: The skill provides example mosquitto_sub and mosquitto_pub bash commands for manual testing. These are well-known, legitimate diagnostic tools for MQTT development and are presented as documentation for the user, not as hidden or automated execution sequences.
[CREDENTIALS_UNSAFE]: The skill demonstrates the WithCredentials(username, password) pattern using variable placeholders, which is a safe practice. It does not contain any hardcoded secrets or API keys.

mqttnet