debug-openshell-cluster
Pass
Audited by Gen Agent Trust Hub on Nov 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONOBFUSCATION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use a wide array of command-line tools for system and cluster diagnostics, including
kubectl,helm,docker,podman,openshell,journalctl,systemctl,rg,stat, andopenssl. - [PRIVILEGE_ESCALATION]: Specific troubleshooting steps require the use of
sudoto perform administrative actions, such as enabling or disabling systemd services and inspecting file metadata in restricted system directories. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The agent is instructed to read and analyze data from untrusted sources, including container logs (
docker logs,kubectl logs), system journals (journalctl), and cluster events. - Boundary markers: The instructions do not include boundary markers or specific warnings to ignore potentially malicious instructions embedded within the logs being analyzed.
- Capability inventory: The skill leverages tools with significant system and cluster-level authority, such as
kubectlfor resource manipulation andsystemctlfor service management. - Sanitization: There is no requirement for the agent to sanitize or validate the content of the logs before interpreting them, creating a surface for indirect prompt injection.
- [OBFUSCATION]: The skill uses hex escapes (
\x1b) within asedcommand intended to strip ANSI color codes from terminal output. While appropriate for the task of log cleaning, this involves the use of encoded character representations.
Audit Metadata