soia-dev-terminal-ops
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to manage shell processes and tmux sessions. It explicitly instructs the agent to perform validation on session names and to obtain user confirmation before executing high-impact commands like
kill -TERM,kill -KILL, or user-provided fallback scripts. It also advises against direct shell string concatenation to prevent command injection. - [EXTERNAL_DOWNLOADS]: The documentation includes an installation command using
npxto add resources from the vendor's repository (soia-team/soia-open-dev-coding-skills). It identifies standard system utilities such astmux,ps,kill, andlsofas dependencies. - [DATA_EXFILTRATION]: No patterns indicative of unauthorized data access or exfiltration were found. Network activity is limited to local process diagnostics via
lsof. - [PROMPT_INJECTION]: The instructions do not contain attempts to override core safety guidelines or agent behavior. It emphasizes adherence to user-defined parameters and confirmation steps.
- [SAFE]: The skill follows security best practices by implementing structured log management, distinguishing between temporary and audit logs, and requiring verification of process IDs to avoid accidental termination of the wrong process.
Audit Metadata