simple-deployment-on-vm
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- Indirect Prompt Injection (LOW): The skill instructs the agent to monitor application logs and perform web requests for health checks. This introduces a surface where malicious content in logs or HTTP responses could influence the agent's subsequent actions.
- Ingestion points: Application logs and HTTP health check status/body (SKILL.md).
- Boundary markers: Absent; no instructions provided to ignore or escape instructions within logs or web responses.
- Capability inventory: Full access to cloud CLI tools (AWS/EC2), file writing (storing passwords/SSH keys), and network operations (health checks).
- Sanitization: Absent; no mention of sanitizing log content or response data before processing.
- Command Execution (SAFE): The skill requires the use of standard CLI tools (cloud provider CLIs, openssl) for deployment and security configuration, which is consistent with its stated purpose.
Audit Metadata