The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform several shell-level operations, including creating directories with mkdir -p, managing files via git mv and git commit, and running a linter with pnpm ai-hero-cli internal lint.
[PROMPT_INJECTION]: The skill is designed to parse and act upon a user-provided 'plan' for scaffolding exercises, creating a surface for indirect prompt injection.
Ingestion points: Untrusted data enters the context when the agent parses a user's plan to extract directory and file names (e.g., SKILL.md workflow step 1).
Boundary markers: The instructions do not specify any delimiters or safety warnings to prevent the agent from obeying instructions that might be embedded within the exercise names or descriptions in the plan.
Capability inventory: The agent has the ability to write files (readme.md), create directories (mkdir), and execute shell commands (git, pnpm).
Sanitization: There are no requirements to validate or sanitize the exercise names extracted from the plan before they are used in filesystem operations or command-line arguments.

scaffold-exercises