NYC

aws-strands

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [Unverifiable Dependencies] (MEDIUM): The skill instructs users to install external packages (strands-agents, strands-agents-tools, and @strands-agents/sdk) that are not from the list of trusted GitHub organizations or repositories. Use of unverified third-party libraries can lead to supply chain attacks.\n- [Indirect Prompt Injection] (LOW): The skill facilitates the creation of agents that process untrusted user input, which could contain malicious instructions.\n
  • Ingestion points: User input passed to agent() or agent.run() in SKILL.md.\n
  • Boundary markers: None present in the provided examples to isolate user input from system instructions.\n
  • Capability inventory: The agents utilize tools (like search_database) and AWS services via boto3, providing a significant impact surface if the agent is compromised.\n
  • Sanitization: No input sanitization or validation is demonstrated in the implementation snippets.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 05:20 PM