The Agent Skills Directory

[SAFE]: The skill provides structured instructions for authentication setup that prioritize security best practices. It explicitly warns against hardcoding secrets, storing tokens in insecure locations, or using over-privileged application permissions.
[SAFE]: External references are restricted to official Microsoft Graph documentation domains. The skill includes specific guidance to treat external service behavior as environment-specific and to "fail closed" if documentation conflicts with observed API responses.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a standard ingestion surface as it processes external documentation and user-provided configuration fixtures.
Ingestion points: External documentation links in references/official-docs.md and user-provided "non-production fixtures" mentioned in SKILL.md.
Boundary markers: The instructions mandate using synthetic data and failing closed when generic metadata conflicts with service-specific statements.
Capability inventory: The skill is authorized to use Write and Edit tools for repository artifacts based on the allowed-tools configuration.
Sanitization: The skill explicitly requires that all fixtures have secrets and real notebook content removed before being processed by the agent.

onenote-install-auth