Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only skill for designing multi-agent roles, with broad example routing rules that users should tighten before using in a real system.
Review the sample routing rules before adopting them. The skill itself is documentation-only, but direct-message catchalls and common keywords such as design, report, test, or bug should be narrowed in real deployments to avoid accidental agent activation or context being sent to the wrong role.
The binding routes any direct-message interaction to the marketing strategist agent without meaningful scoping, making agent selection overly permissive. In a multi-agent environment, this can cause unintended invocation, misrouting of sensitive user requests, or broaden the strategist agent’s exposure to inputs it was not intended to handle.
The example bindings use highly generic keywords like content, design, report, test, and bug, which are common in ordinary conversation and likely to collide with unrelated requests. This can lead to accidental agent activation, incorrect routing, and context leakage between roles if user messages are dispatched to agents based on ambiguous terms alone.
No suspicious patterns detected.
