The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to automate development tasks:
Executes mkdir -p to create directory structures for exercises based on user-provided plans.
Uses git mv and git commit for version control management of the scaffolded content to maintain history.
Runs pnpm ai-hero-cli internal lint to validate the created structure against project-specific rules.
[INDIRECT_PROMPT_INJECTION]: The skill parses an external 'plan' to determine directory names and file content, which are then used as arguments in shell commands. This creates a surface for indirect prompt injection if the plan contains malicious strings designed to escape shell commands.
Ingestion points: The instructions specify parsing a 'plan' (e.g., in the 'Example: stubbing from a plan' section) to extract section and exercise names.
Boundary markers: None identified; the skill directly interpolates extracted values into command strings.
Capability inventory: File system modification (mkdir), version control operations (git), and execution of a specific CLI tool (pnpm ai-hero-cli).
Sanitization: The skill includes explicit instructions to the agent to use 'dash-case' (lowercase, hyphens), which acts as a logical constraint on the generated paths, though it relies on the LLM's adherence to the instructions.

scaffold-exercises