microsoft-development

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external documentation from Microsoft Learn using specialized tools. This establishes a surface for indirect prompt injection where instructions embedded in external content could potentially influence the agent's behavior.
  • Ingestion points: Content retrieved from the web via microsoft_docs_fetch and microsoft_docs_search tools defined in SKILL.md.
  • Boundary markers: The instructions do not define specific delimiters or guardrails to isolate external documentation content from the agent's primary command flow.
  • Capability inventory: The skill provides access to local execution environments through the scripts/azure-health-check.ps1 script and documentation for various command-line tools (az, dotnet, npm).
  • Sanitization: No explicit sanitization or filtering of the fetched documentation content is implemented.
  • [CREDENTIALS_UNSAFE]: A hardcoded credential string is present in a configuration file within the examples directory.
  • Evidence: The file examples/azure-function-api-example.md includes a local.settings.json sample with a hardcoded AccountKey for Azure Cosmos DB.
  • Context: The provided key (C2y6yDjf5/R+ob0N8A7Cgv30VRDJIWEHLM+4QDU5DE2nQ9nDuVTqobD4b8mGGyPMbIZnqyMsEcaGQy67XIw/Jw==) is the standard, publicly documented default key for the Azure Cosmos DB Emulator. While it triggers credential detection patterns, it is a non-sensitive value intended for local development environments.
  • [EXTERNAL_DOWNLOADS]: The skill references and provides installation instructions for an external Model Context Protocol (MCP) server.
  • Evidence: SKILL.md and references/microsoft-learn-mcp.md suggest the installation of @microsoft/learn-docs-mcp via npx.
  • [COMMAND_EXECUTION]: The skill includes a utility script intended for local execution to verify the status of cloud resources.
  • Evidence: The PowerShell script scripts/azure-health-check.ps1 executes several Azure CLI (az) commands to inspect Web Apps, Function Apps, and databases within a specified resource group.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 06:18 AM