Security audit
Security checks for vulnerabilities and agentic risk
This skill is a straightforward Halo CLI guide for comment moderation and notifications, with destructive delete commands disclosed and aligned to that purpose.
Install only for agents you trust to moderate Halo content. Treat delete commands with --force as intentional destructive actions, verify the selected profile and resource ID first, and prefer mark-as-read or review/get commands when deletion is not explicitly requested.
The skill documents halo comment delete ... --force as a common flow but does not pair it with an explicit warning that the action is destructive and may be irreversible. In an agent-facing skill, normalizing forced deletion without guardrails can lead to accidental data loss or unsafe automation, especially if an agent follows examples verbatim.
The notification deletion example uses halo notification delete ... --force without clearly warning about the data-impacting consequences. In this moderation/notification context, agents may treat documented commands as approved defaults, increasing the risk of silently deleting records instead of marking them read or otherwise preserving auditability.
No suspicious patterns detected.
