The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands for project management and validation, including mkdir -p, git mv, git commit, and pnpm ai-hero-cli internal lint. These commands are used to create directory structures and verify them against local project rules.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it instructs the agent to parse a user-supplied 'plan' and use extracted strings as arguments for shell commands (e.g., directory names for mkdir).
Ingestion points: The agent parses a user-provided 'plan' to extract section and exercise names (SKILL.md).
Boundary markers: There are no explicit instructions to use delimiters or sanitization when interpolating the plan content into shell commands.
Capability inventory: The skill possesses filesystem write capabilities (mkdir), version control manipulation (git), and local command execution (pnpm).
Sanitization: The instructions do not specify any validation or escaping for the strings extracted from the user's plan before they are used in shell executions.

scaffold-exercises