The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several shell commands including mkdir, git mv, and git commit to manage the exercise directory structure and version control. It also executes pnpm ai-hero-cli internal lint to validate the created content, which involves running a project-specific executable within the workspace environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and parse a "plan" provided by the user to determine directory and file names. This creates an attack surface where malicious input in the plan could attempt to influence the agent's file system operations, although the skill instructions include strict naming conventions and structural rules that act as constraints.
Ingestion points: User-provided exercise plan described in SKILL.md workflow.
Boundary markers: None identified.
Capability inventory: mkdir, git mv, git commit, file writing (readme.md), and pnpm command execution.
Sanitization: No explicit sanitization of the input plan content is mentioned before it is used to generate shell commands and directory names.

scaffold-exercises