realtime-cinema-data-engineering-pipeline

Warn

Audited by Gen Agent Trust Hub on May 25, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The installation guide instructs users to clone a repository from an untrusted third-party GitHub account (BaidaneAyoub/realtime-cinema-data-engineering.git). \n- [COMMAND_EXECUTION]: The skill provides instructions to execute several shell commands, including git clone, python environment setup, pip dependency installation, and the execution of the downloaded Python scripts (main_producer.py, main_consumer.py). \n- [CREDENTIALS_UNSAFE]: The documentation and code snippets use default credentials (e.g., 'postgres:postgres') for database access. While these serve as placeholders in a tutorial context, they represent insecure defaults. \n- [PROMPT_INJECTION]: The skill implements an architecture that processes external, untrusted JSON data from a Kafka stream, creating a surface for indirect prompt injection. \n
  • Ingestion points: Untrusted data enters the pipeline via the KafkaConsumer in 'consumer/main_consumer.py'. \n
  • Boundary markers: None identified in the Kafka event processing logic. \n
  • Capability inventory: The skill has the capability to perform database writes (PostgreSQL) and orchestrate tasks via Apache Airflow. \n
  • Sanitization: The code uses parameterized SQL queries (psycopg2) which protects against SQL injection, but the system remains vulnerable to logical indirect injection via the processed JSON payloads.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 25, 2026, 01:18 AM
Security Audit — agent-trust-hub — realtime-cinema-data-engineering-pipeline