The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and process documentation from external URLs, which creates an attack surface for indirect prompt injection if the source content is compromised.
Ingestion points: Documentation is retrieved from learn.microsoft.com via the mcp_microsoftdocs:microsoft_docs_fetch or fetch_webpage tools.
Boundary markers: The instructions do not define delimiters or specific boundary markers to isolate external content from the agent's instructions.
Capability inventory: The skill utilizes network-enabled tools to fetch markdown and text content.
Sanitization: There are no explicit instructions for the agent to sanitize or ignore embedded instructions within the fetched documentation.
[EXTERNAL_DOWNLOADS]: The skill contains numerous references to external resources for documentation and tool installation.
Source URL: Numerous links to learn.microsoft.com for technical guidance.
Source URL: Links to github.com/MicrosoftDocs/mcp for tool installation instructions.
Analysis: These references point to official vendor infrastructure (Microsoft) and are considered safe for the intended documentation-retrieval purpose.

azure-web-pubsub