The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes pnpm ai-hero-cli internal lint to validate the directory structure after creation. It also performs version control operations using git commit and git mv, and modifies the filesystem using mkdir -p.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided exercise plans to determine directory and file names, creating an attack surface where untrusted input could attempt to influence filesystem operations.
Ingestion points: Exercise plans, section names, and titles provided by the user as described in SKILL.md.
Boundary markers: The instructions specify a strict "dash-case" (lowercase, hyphens) naming convention, which serves as a constraint on the generated paths.
Capability inventory: The skill possesses capabilities to write to the filesystem (mkdir), execute local scripts (pnpm), and perform git operations.
Sanitization: The requirement to transform user-provided names into "dash-case" provides a layer of implicit sanitization before names are used in shell commands or path strings.

scaffold-exercises