Security risk report for @forgecat/microsoft-azure_skills-azure-aigateway v0.0.5
Source Integrity
Low
Profile is attributed to Microsoft and references official Microsoft documentation (learn.microsoft.com), Azure SDK repositories, and Azure-Samples GitHub org.
Static scanner flagged low risk; source appears legitimate and well-established, with clear attribution and links to official Microsoft resources.
Agent Intent
Low
Content describes legitimate Azure API Management configuration patterns for AI governance (caching, rate limiting, content safety) without instructing the AI to ignore instructions, exfiltrate data, or manipulate its own behavior.
Documentation includes security best practices (managed identity, RBAC, least privilege) and does not plant malicious guidance or weaken security defaults; recommendations align with Microsoft's official security guidance.
No evidence of prompt injection, system prompt leakage, or guidance poisoning; the profile educates on proper AI gateway setup rather than subverting it.
Details
Evidence
auth-best-practices.md explicitly recommends ManagedIdentityCredential in production and warns against DefaultAzureCredential in production environments.
policies.md documents content safety, jailbreak detection, and rate limiting as protective measures, not as bypasses.
No instructions to read credentials, exfiltrate data, or hide the profile's own instructions.
Permissions
Low
Profile declares no tools, hooks, or MCP servers; it is purely documentation and configuration guidance.
Content describes Azure CLI commands and SDK patterns for legitimate API Management administration, all scoped to the user's own Azure resources.
No excessive agency or unrelated high-risk permissions requested.
Details
Evidence
All bash examples use `az apim` and `az cognitiveservices` commands scoped to named resource groups and subscriptions.
SDK code examples (Python, TypeScript, .NET, Java) use standard Azure Identity libraries with appropriate credential selection.
No file_write, file_delete, shell execution, or subagent directives in the profile.
MCP Risk
Low
Profile declares no MCP servers; it is documentation and configuration guidance only.
References to Azure services (Azure OpenAI, Content Safety, API Management) are external dependencies, not embedded MCP definitions.
No hidden instructions, arbitrary binary execution, or unrestricted network/filesystem access in the profile itself.
Details
Evidence
No MCP server definitions in the profile.
External service references (Azure OpenAI, Content Safety) are documented as prerequisites, not as embedded or hidden MCP payloads.
ForgeCat