wiz-security-context
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- Dynamic Instruction Loading: The skill functions as a stub that retrieves its primary logic (a 'workflow') from a remote Wiz MCP server (mcp.app.wiz.io). This design allows the auditing workflow to be updated by the service provider without requiring changes to the skill itself. The agent follows these instructions at runtime, which constitutes a dynamic execution pattern.
- Indirect Prompt Injection Surface: The agent's decision-making is influenced by content returned from the devops_resource_auditing_skill tool. This creates a surface where external content could potentially direct the agent's behavior. This risk is managed by the requirement for users to explicitly allowlist and authenticate the Wiz MCP server. * Ingestion points: Data received from the remote devops_resource_auditing_skill tool in SKILL.md. * Boundary markers: Explicit markers or 'ignore' instructions for the tool output are not present. * Capability inventory: The skill is scoped to read-only queries of security posture via Wiz tools. * Sanitization: No specific sanitization process is described for the incoming workflow data before the agent processes it.
Audit Metadata