aws-review

Warn

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The discovery process in SKILL.md (Step 1) explicitly directs the agent to locate and record the .aws/credentials file. Accessing this sensitive file path poses a risk of exposing AWS access keys and secrets within the agent context.
  • [PROMPT_INJECTION]: The skill processes untrusted infrastructure-as-code and configuration files, creating a surface for indirect prompt injection.
  • Ingestion points: The skill reads various configuration files including Terraform (.tf), CloudFormation (.yaml, .json), and CDK source files.
  • Boundary markers: SKILL.md contains a Prompt Injection Safety Notice that instructs the agent to treat configuration values as data rather than instructions.
  • Capability inventory: The agent uses Read, Grep, and Glob tools to inspect files throughout the assessment process.
  • Sanitization: The skill relies on natural language instructions for the agent rather than technical sanitization of the file contents.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 26, 2026, 01:49 PM
Security Audit — agent-trust-hub — aws-review