test-environment-management
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION] (HIGH): High risk of indirect prompt injection via agent task parameters.
- Ingestion points: Data passed to
Task("Environment Provisioning", ...)andFleetManager.coordinate(specifically theservices,baseline, andstrategyfields) inSKILL.md. - Boundary markers: Absent. The skill provides no instructions to treat task parameters as data only, nor does it use delimiters to separate instructions from untrusted input.
- Capability inventory: Direct execution of infrastructure-altering tools including AWS CLI (
aws ec2 stop-instances/start-instances), Docker Compose (docker-compose up/exec), and Terraform provisioning. - Sanitization: Absent. No validation or filtering is specified for the environment names, service lists, or tags used in the automated scripts.
- [COMMAND_EXECUTION] (MEDIUM): The skill utilizes high-privilege shell commands for infrastructure management.
- Evidence: Use of
aws ec2with query interpolation$(aws ec2 describe-instances ...)anddocker-compose execcan be dangerous if the underlying filters or service names are manipulated. - [EXTERNAL_DOWNLOADS] (LOW): The skill references external dependencies.
- Evidence: Requires the
wiremock-captainnpm package and pulls Docker images (postgres:15,redis:7). These are standard dependencies but increase the attack surface via the supply chain.
Recommendations
- AI detected serious security threats
Audit Metadata