The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill directs the agent to fetch documentation from learn.microsoft.com. This is the official, well-known domain for Microsoft's technical documentation and is considered a trusted source.
[EXTERNAL_DOWNLOADS]: A reference to an installation guide for an MCP tool is provided, pointing to the MicrosoftDocs GitHub organization. As MicrosoftDocs is a trusted organization and the author of the skill, this reference to their official repository for configuration purposes is considered safe.
[INDIRECT_PROMPT_INJECTION]: The skill identifies an ingestion surface as it retrieves external documentation content at runtime. However, the risk is mitigated by the trusted nature of the source and the lack of dangerous capabilities within the skill's instructions.
Ingestion points: External documentation fetched from learn.microsoft.com using the mcp_microsoftdocs:microsoft_docs_fetch tool or fetch_webpage fallback.
Boundary markers: Not explicitly defined in the instruction set for the fetched content.
Capability inventory: The skill does not possess or invoke tools for file-system writing, arbitrary command execution (shell/subprocess), or sending data to third-party network endpoints.
Sanitization: No specific content filtering or sanitization logic is implemented for the fetched documentation strings.

azure-databricks