The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches documentation from 'learn.microsoft.com' and provides links to 'github.com/MicrosoftDocs'. These are well-known services and official resources belonging to the vendor, which are considered safe sources.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes content from external web sources, which is a common pattern for documentation agents but introduces a potential surface for instructions embedded in remote data.
Ingestion points: Documentation is retrieved at runtime using the mcp_microsoftdocs:microsoft_docs_fetch and fetch_webpage tools as defined in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following instructions that might be present in the fetched documentation.
Capability inventory: The skill utilizes tools to fetch and read external markdown content and provides guidance on authoring Azure Policy JSON as seen in SKILL.md.
Sanitization: No specific sanitization or validation logic for the external markdown content is mentioned in the skill instructions.

azure-policy