data-engineering

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The documentation includes a command to install the neo4j-cli tool using curl piped to bash from the official neo4j.sh domain. This is a common installation method for well-known developer tools.
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent on how to fetch and install various dependencies, including official dbt packages (e.g., dbt-core, dbt-labs/dbt_utils) and community packages (e.g., calogica/dbt_expectations). It also references official plugins and drivers for database systems like Neo4j and InfluxDB.
  • [INDIRECT_PROMPT_INJECTION]: As the skill focus is on building pipelines that ingest raw data from external sources, it naturally describes an attack surface for indirect prompt injection. The documentation mitigates this risk by recommending extensive data quality validation, schema enforcement, and boundary checks throughout the ETL process.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 01:03 PM
Security Audit — agent-trust-hub — data-engineering