datadog

Fail

Audited by Gen Agent Trust Hub on Oct 4, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill instructs users to download and execute the official Datadog Agent installation script from Amazon S3 using a subshell execution pattern (bash -c "$(curl ...)").
  • [EXTERNAL_DOWNLOADS]: Downloads official Datadog packages, libraries, and Helm charts from standard repositories and registries (APT, PyPI, npm, and helm.datadoghq.com).
  • [COMMAND_EXECUTION]: Instructions include the use of various system and deployment tools such as systemctl, kubectl, helm, and git to manage the monitoring agent and environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill configuration creates a potential surface for indirect prompt injection by enabling the ingestion of application-generated logs.
  • Ingestion points: Log files located at /var/log/myapp/*.log (SKILL.md).
  • Boundary markers: No delimiters or instructions were identified to distinguish between log data and agent instructions.
  • Capability inventory: The skill provides instructions that utilize systemctl, apt-get, helm, npm, pip, and kubectl (SKILL.md).
  • Sanitization: There are no provided log processing rules specifically designed to filter or sanitize executable instructions within the ingested log content.
Recommendations
  • HIGH: Downloads and executes remote code from: https://s3.amazonaws.com/dd-agent/scripts/install_script_agent7.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Oct 4, 2026, 08:51 AM