loopback_users = none in rabbitmq.conf lets guest/guest log in from any host, and listeners=PLAINTEXT://:9092 in a Kafka server.properties with no authorizer lets anyone who can reach the port read, write and delete topics. This skill finds those and six other kinds of message broker mistakes before the broker goes live, and tells you how to fix each one.

Your agent runs a scanner on your broker configs: RabbitMQ (rabbitmq.conf, conf.d files, advanced.config), Mosquitto (mosquitto.conf), Kafka (server.properties, broker.properties, controller.properties) and NATS (nats.conf, nats-server.conf), plus .conf files in a conf.d folder or next to those files (such as /etc/mosquitto/conf.d/default.conf), whose format it works out from the content. It reads each format the way the broker does (Erlang terms, per-listener Mosquitto settings, Java properties with line continuation, NATS blocks and arrays) and knows each broker's defaults: RabbitMQ listens on every interface and keeps guest on localhost, Mosquitto 2.x refuses anonymous clients once a listener is defined, an empty Kafka listener host binds every interface and KRaft controller listeners default to PLAINTEXT, and NATS listens on 0.0.0.0. It never connects to a broker. It returns BLOCK, REVIEW, or NOT FLAGGED, with the file, line, reason, and a safer approach. Secrets are never printed.

What it checks

BLOCK (fix before shipping)

• B01: clients let in without credentials on a network listener (Mosquitto allow_anonymous true, Kafka PLAINTEXT or SSL without client certificates and no authorizer, NATS with no authorization or with no_auth_user, RabbitMQ anonymous_login_user)

• B02: the RabbitMQ guest user allowed from remote hosts, or default_pass set to guest, the user name or a common word

• B03: Kafka allow.everyone.if.no.acl.found=true

• B04: passwords, tokens and keys written in the config (RabbitMQ default_pass and ssl_options.password, Mosquitto bridge remote_password, Kafka keystore passwords, inline private keys and sasl.jaas.config, NATS passwords, tokens and route URLs)

REVIEW (check before release)

• B11: listeners reachable from the network without TLS, including plain RabbitMQ 5672 left open next to a TLS listener

• B12: the RabbitMQ management UI (HTTP or HTTPS) or the NATS monitoring port on every interface

• B13: Kafka topic auto-creation on (auto.create.topics.enable=true, or not set)

• B14: NATS cluster or gateway ports without authorization

What's included

• The scanner (Python 3.9+, standard library only)

• A rules guide with the reason and safer approach for each rule

• A risky and a safe example project (RabbitMQ, Mosquitto, Kafka and NATS configs), and the real report the scanner produces on the risky one

Limits

It reads each file on its own: settings in one conf.d or include file are not combined with another, and environment variables and command-line flags (Docker RABBITMQ_* and KAFKA_* variables), password and ACL files, Kafka ACLs stored in the cluster, enabled plugins and firewalls are not checked. Mosquitto rules assume version 2.x. NOT FLAGGED is not a guarantee.

Contents

One ZIP file with 14 files: SKILL.md, the scanner in scripts/ (3 Python files), references/rules.md (every rule with its reason and safer approach), and examples/ (8 example files plus report.md, the real scanner output on the risky example).

How to use: This is an agent skill: a folder with a SKILL.md file and Python scripts that an AI coding agent such as Claude Code, Cursor, or Codex CLI can run. Unzip it into your agent's skills folder (for Claude Code: ~/.claude/skills/), then ask your agent to check your project. The scripts also run on their own with Python 3.9 or newer, with nothing to install. They read your files only: no network access, and nothing is sent anywhere.