microsoft-skill-creator

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill utilizes and recommends the @microsoft/learn-cli package, which is an official tool from Microsoft, via npx or global npm installation.
  • [INDIRECT_PROMPT_INJECTION]: The skill functions as a generator for other skills by ingesting content from Microsoft Learn, representing an attack surface for instructions embedded in external documentation.
  • Ingestion points: Remote documentation content is ingested through the microsoft_docs_fetch and microsoft_docs_search tools as defined in the investigation phase.
  • Boundary markers: The skill templates provided do not implement specific boundary markers to isolate remote fetched content from the agent's instructions.
  • Capability inventory: The skill has the capability to scaffold new directory structures and files, and it encourages the use of shell commands for tool execution.
  • Sanitization: The skill assumes the safety of content retrieved from the official learn.microsoft.com domain and does not include explicit data sanitization logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 09:00 AM