vpc-air-gapped
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Documentation instructs the user to fetch the Rivet Engine source code from the official GitHub repository and pull container images from the 'rivetdev' organization on Docker Hub.
- [COMMAND_EXECUTION]: Provides detailed guidance for executing system commands including compilation with 'cargo build', container management with 'docker', and service control with 'systemctl'. It also covers shell and process management within virtualized 'agentOS' environments.
- [PROMPT_INJECTION]: The skill describes patterns for AI agents that ingest untrusted external data (e.g., Slack webhooks, chat messages) while possessing capabilities like filesystem access and shell execution. This creates a surface for indirect prompt injection where malicious data processed by the agent could influence its actions.
Audit Metadata