azure-functions-connector-mcp
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- Command Execution: The skill utilizes
az restand other Azure CLI commands to interact with Azure Management APIs. This is a standard and expected method for managing cloud infrastructure and is used here to discover connector operations, export Swagger definitions, and register MCP configurations. - Indirect Prompt Injection Surface: The skill processes connector metadata from Azure platform services to generate tool definitions, which is an inherent part of its functionality.
- Ingestion points: Metadata is retrieved from
apiOperationsand Swagger exports as detailed inreferences/metadata-to-mcp.md. - Boundary markers: The skill uses structured JSON formats for configuration registration and manual selection of operations.
- Capability inventory: The skill uses
az restfor network communication and resource management as seen inSKILL.mdandreferences/metadata-to-mcp.md. - Sanitization: The instructions emphasize using authoritative metadata to ensure schema accuracy and reliability, reducing the risk of processing unintended instructions.
- Sensitive Data Management: The workflow involves handling Azure subscription IDs and generated endpoints. The skill provides clear guidelines to redact these values from logs and manage them via environment-scoped variables or app settings instead of committing them to source control, which aligns with industry security best practices.
Audit Metadata