datadog-design-dashboard

Warn

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Shell tool to execute pup CLI commands, instructing the agent to interpolate user-provided values such as dashboard IDs and service names directly into command strings (e.g., pup dashboards get <dashboard-id> --agent). This pattern creates a command injection vulnerability if the inputs are not properly sanitized.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it reads and processes external dashboard definitions and codebase contents. An attacker could embed malicious instructions in these data sources to influence the agent's behavior during audits or designs.\n
  • Ingestion points: Dashboard JSON returned by pup and file contents accessed via the Read tool.\n
  • Boundary markers: None; the instructions do not specify delimiters or safety warnings for external content.\n
  • Capability inventory: The skill uses Shell, Write, and Read tools, allowing for file system and command-line interactions.\n
  • Sanitization: No validation or sanitization of external data or user-supplied parameters is described.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 20, 2026, 05:11 PM
Security Audit — agent-trust-hub — datadog-design-dashboard