The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to use mkdir -p and git mv for managing exercise directories. These operations are restricted to the local filesystem and are intended for project organization.
[COMMAND_EXECUTION]: The skill encourages running existing repository validation or linting commands. This is a standard practice for ensuring structural integrity within a development environment.
[INDIRECT_PROMPT_INJECTION]: The skill parses a user-provided plan to determine the names and types of exercises to create. This represents an ingestion point for untrusted data.
Ingestion points: User-provided exercise plan (SKILL.md).
Boundary markers: None explicitly defined for the input plan.
Capability inventory: Uses mkdir, git mv, and file writing (readme.md) (SKILL.md).
Sanitization: The skill enforces strict naming conventions (e.g., numeric prefixes, dash-case) and requires a validation step to ensure the resulting structure meets repository standards, which serves as a mitigation against malicious path manipulation.

scaffold-exercises