kafka-attack
Pass
Audited by Gen Agent Trust Hub on Aug 12, 2026
Risk Level: SAFECREDENTIALS_UNSAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill describes methods to search for and extract sensitive information like passwords, secrets, tokens, and AWS access keys from Kafka and Kafka Connect configurations using tools like grep and jq.
- [DATA_EXFILTRATION]: The skill provides instructions for dumping large amounts of message data from Kafka topics to local files in the /tmp directory, and instructions for using Kafka Connect connectors to read sensitive local system files like /etc/passwd and pipe the content to an external topic.
- [COMMAND_EXECUTION]: The skill uses multiple system-level commands and external tools including nmap, kcat, zkCli, and various Kafka shell scripts to perform its tasks.
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it reads and processes data from external Kafka topics without validation. Ingestion points: Message data consumed in SKILL.md and references/attack-techniques.md. Boundary markers: Absent. Capability inventory: Subprocess calls and file writing in SKILL.md. Sanitization: Absent.
Audit Metadata