enable-dsm
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes administrative commands using kubectl and helm to update cluster configurations, and ssh to manage Linux host services.\n- [PRIVILEGE_ESCALATION]: The skill uses sudo on Linux systems to modify systemd service environments and restart applications. These actions are performed only after obtaining explicit user consent as directed by the instructions.\n- [INDIRECT_PROMPT_INJECTION]: The skill discovers messaging clients by scanning project files and cluster metadata, which introduces a potential injection surface.\n
- Ingestion points: Discovery relies on grep output from local repository files and pod metadata retrieved through kubectl (SKILL.md).\n
- Boundary markers: The instructions explicitly mandate that the agent treat all detection results strictly as data and disregard any scripts or instructions found within repository files or cluster metadata.\n
- Capability inventory: The skill includes capabilities to apply Kubernetes manifests, upgrade helm releases, execute commands via SSH, and query the Datadog API (SKILL.md).\n
- Sanitization: Safety relies on instructional constraints that prevent the agent from executing content retrieved during discovery, ensuring it is only used to evaluate the service's fit for monitoring.
Audit Metadata