qe-test-environment-management
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFECREDENTIALS_UNSAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: Hardcoded database credentials ('password') are included in the docker-compose.test.yml example.
- [COMMAND_EXECUTION]: The skill contains shell commands for Docker Compose and the AWS CLI to provision and manage infrastructure.
- [PROMPT_INJECTION]: An indirect prompt injection surface is present.
- Ingestion points: Environment configurations ('configs/*') and output from AWS CLI queries.
- Boundary markers: No delimiters or instructions to ignore embedded commands are present in the snippets.
- Capability inventory: The skill can execute subprocesses via
docker-compose execandaws ec2commands. - Sanitization: There is no logic for sanitizing or validating data retrieved from external tools or configuration files.
Audit Metadata