Security audit
Security checks for vulnerabilities and agentic risk
This AWS guidance skill is mostly purpose-aligned, but its safety and privacy statements understate real AWS CLI effects and some triggers are too broad.
Install only if you want AWS-focused guidance and are prepared to review every AWS CLI command before execution. Treat examples as advisory, not automatically safe; verify whether each command reads, creates, modifies, deletes, or can incur cost, and avoid enabling this skill for generic web or deployment conversations unless AWS is explicitly intended.
The trigger keywords include very broad terms such as 'web', 'deploy', 'architect', and 'optimize', which can cause the skill to activate in unrelated conversations. In an agentic environment, over-broad activation increases the chance that AWS-oriented guidance or command suggestions are injected into contexts where the user did not intend cloud operations.
The repeated trigger section reinforces ambiguous activation behavior without additional guardrails, again using generic terms that overlap with normal conversation. Because this skill can suggest executable cloud commands, unintended invocation is more dangerous than in a purely informational skill.
The file states that no data leaves the machine, but AWS CLI commands inherently transmit request metadata and potentially account or resource information to AWS APIs. This is misleading security/privacy documentation that may cause users to underestimate data exposure when using the skill.
The skill claims that all shown CLI commands are read-only by default and that destructive actions require explicit confirmation, but the document includes modifying commands such as creating VPC endpoints and setting log retention policies. This mismatch can mislead users or an agent into executing state-changing AWS actions under a false assumption of safety, increasing the risk of unintended infrastructure changes and cost impact.
No suspicious patterns detected.
