deploying-scalable-agents
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
- [External Reference]: The skill references a GitHub Action ('ai-smoke-test') from the GitHub Marketplace to automate post-deployment health checks. This is a standard operational practice for verifying that agent endpoints are responding correctly after a deployment.
- [Access Control & Identity]: The instructions explicitly advocate for migrating from personal login credentials to Managed Identities and scoped RBAC (Role-Based Access Control) in production environments, which aligns with industry best practices for least-privilege access.
- [Safety & Human Oversight]: It includes patterns for high-risk actions, such as large refunds, that require explicit human approval via tools. This provides a critical safeguard against autonomous errors in production.
- [Boundary Security]: The skill identifies external components, specifically Model Context Protocol (MCP) servers, as untrusted boundaries. It recommends version pinning, identity scoping, and output validation to manage the security surface of these integrations.
Audit Metadata