The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches the mqtt-pwn framework from Akamai Threat Research's official GitHub repository (github.com/akamai-threat-research/mqtt-pwn).
[COMMAND_EXECUTION]: The skill documents the use of shell commands for environment setup, including docker-compose and pip install, as well as interactive shell commands like exec <cmd> for remote command execution on targets during command-and-control (C2) workflows.
[DATA_EXFILTRATION]: The tool is designed to identify and extract sensitive information from IoT devices and MQTT brokers, including device fingerprints, Wi-Fi credentials, and broker authentication secrets.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, untrusted MQTT brokers which could contain malicious payloads targeting the analysis environment.
Ingestion points: Processes messages and topic names gathered during the discovery and messages phases (SKILL.md).
Boundary markers: No specific boundary markers or sanitization logic for ingested message content are described in the instructions.
Capability inventory: The tool includes high-privilege capabilities such as shell command execution (exec) and persistent storage of captured data in a PostgreSQL database (SKILL.md).
Sanitization: The documentation does not specify sanitization or validation routines for data received from the broker before it is processed or stored.

mqtt-pwn